UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Estimation of limb occlusion pressure for surgical tourniquets based on the measurement of arterial pulse… Marko, Alexei John 1994

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_1995-0023.pdf [ 9.96MB ]
Metadata
JSON: 831-1.0065135.json
JSON-LD: 831-1.0065135-ld.json
RDF/XML (Pretty): 831-1.0065135-rdf.xml
RDF/JSON: 831-1.0065135-rdf.json
Turtle: 831-1.0065135-turtle.txt
N-Triples: 831-1.0065135-rdf-ntriples.txt
Original Record: 831-1.0065135-source.json
Full Text
831-1.0065135-fulltext.txt
Citation
831-1.0065135.ris

Full Text

ESTIMATION OF LIMB OCCLUSION PRESSURE FOR SURGICAL TOURNIQUETS BASED ON THE MEASUREMENT OF ARTERIAL PULSE WAVE TRANSIT TIME by Alexei J o h n Marko B . A . S c . Queen's University, 1990 A THESIS SUBMITTED IN P A R T I A L F U L F I L L M E N T O F T H E R E Q U I R E M E N T S F O R T H E D E G R E E O F M A S T E R OF A P P L I E D S C I E N C E in T H E F A C U L T Y O F G R A D U A T E STUDIES D E P A R T M E N T OF E L E C T R I C A L ENGINEERING We accept this thesis as conforming to the required standard T H E UNIVERSITY OF BRITISH C O L U M B I A November 1994 © Alexei John Marko, 1994 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. The University of British Columbia Vancouver, Canada Department DE-6 (2/88) II ABSTRACT While the use of pneumatic surgical tourniquet systems to maintain a bloodless surgical field in limb surgeries has become commonplace, their use may occasionally result in serious complications such as nerve injury and there is evidence that some injury occurs as a result of each usage. It is commonly accepted that these complications are a result of the pressure applied by the tourniquet cuff to the underlying tissue. The use of an adaptive surgical tourniquet system which adapts to maintain the minimum tourniquet pressure required for surgery may allow the use of tourniquet pressures that are on average much lower than could be achieved using a conventional tourniquet system set at a constant pressure and this lower pressure may result in reduced injury and thus improved safety. Although many variables affect the minimum tourniquet pressure required to prevent the flow of blood underneath the tourniquet cuff, intraoperative changes in that minimum tourniquet pressure are primarily a function of intraoperative changes in blood pressure. Prior to work described in this thesis, no clinically acceptable means had been developed for estimating changes in required tourniquet pressure due to changes in patient blood pressure at sufficiently frequent intervals to facilitate the operation of an adaptive surgical tourniquet system. In the research described in this thesis, the measurement of "pulse wave transit time" (PWTT), defined as the time required for the arterial pulse wave to propagate along a fixed arterial path, was investigated as a novel technique to allow the continuous estimation of "limb occlusion pressure" (LOP) for the control of adaptive surgical tourniquet systems. An Ill experimental system was designed and built which measures PWTT using the output from an ECG monitor and an arterial pulse sensor. Initially, data was collected using the measurement system from 16 patients undergoing orthopedic surgical procedures. Based on analysis of this data it was concluded that in relatively short procedures performed on non-elderly subjects there existed a strong correlation between PWTT and LOP, and therefore it was concluded that measured PWTT provided an accurate method of continuously estimating LOP in these cases. However, in lengthy cases involving elderly patients, a poor correlation was found between PWTT and LOP. It was concluded that this was the result of the relative non-compliance of the arteries of elderly subjects and the administration of a wide variety of drugs affecting arterial compliance. Based on these findings, it was determined that in order to achieve a clinically acceptable level of reliability, a system which continuously estimates LOP based on the measurement of PWTT must integrate periodic estimates from a conventional blood pressure measurement device. These periodic estimates are required in order to develop a patient-specific model relating LOP to PWTT, and to identify subjects for whom PWTT is poorly correlated with LOP. An improved system was then proposed based on the integration of the PWTT measurement system with a patient monitor commonly used in operating rooms to provide blood pressure estimates periodically. Applying the algorithm proposed for the improved system retrospectively to the clinical data obtained from the initial study of 16 surgical patients yielded a 9 percent reduction in the time-averaged tourniquet pressure and a 27 percent average decrease in pressure during the periods of adaptive operation. A follow-up study on 8 additional surgical patients yielded a 12 percent estimated decrease in time-averaged tourniquet pressure and a 25 percent reduction in tourniquet pressure during periods of adaptive operation. TABLE OF CONTENTS iv Abstract ii List of Figures viii List of Tables x Medical Terminology xi Acknowledgments xiv 1 INTRODUCTION 1 1.1 Motivation for the Research 1 1.2 Scope of the Research 5 1.3 Contributions of the Research 6 1.4 Thesis Overview 7 2 BACKGROUND AND REVIEW OF PREVIOUS RESEARCH 9 2.1 Surgical Tourniquets 9 2.1.1 Background 9 2.1.2 Modern pneumatic tourniquets 9 2.1.3 Method of application 10 2.1.4 Safety problems in the use of tourniquets 11 2.2 Limb Occlusion Pressures 12 2.2.1 The selection of tourniquet pressure and factors which affect LOP 12 2.2.2 The use of measured LOP for determining minimum effective tourniquet pressure 17 2.3 The Adaptive Tourniquet 20 2.3.1 Development of an algorithm to allow adaptation of a tourniquet based on measured LOP and the monitoring of intraoperative changes in SBP 20 2.3.2 Clinical requirements of an adaptive surgical tourniquet system 23 2.4 Blood Pressure Measurement Techniques 24 2.4.1 The physiology of blood pressure 24 V 2.4.2 Current technologies for the estimation of blood pressure 26 3 THE ADAPTATION AND EVALUATION OF ALGORITHMS EMPLOYING PULSE WAVE TRANSIT TIMES TO ESTIMATE REQUIRED TOURNIQUET PRESSURE 3.1 The Arterial System 3.1.1 Physiology 3.1.2 Wave reflections 3.1.3 Shape of the pressure pulse 3.2 The Cardiac Cycle and Interpretation of the ECG 3.3 The Theoretical Background for Relating PWTT to BP 3.4 The Development of an Algorithm for Estimating Ideal Tourniquet Pressure Based on Measured PWTT 3.5 The Selection of Slope Coefficients for Relating APWTT to ASBP and the Selection of the Pulse Path for Measurement 4 THE DESIGN AND IMPLEMENTATION OF A PROTOTYPE PWTT MEASUREMENT SYSTEM TO ALLOW ESTIMATION OF LOP FOR TOURNIQUET-RELATED USAGE 50 4.1 System Overview 50 4.2 Detection of the R-wave from the ECG Signal 52 4.3 The Arterial Pulse Sensor 56 4.3.1 Operational requirements 56 4.3.2 Available technologies for pulse sensing 57 4.3.3 The photoplethysmographic sensor 58 4.3.4 The experimental sensor 66 4.3.5 Modification of the sensor for application under the tourniquet cuff 72 4.4 Pre-Conditioning of the Pulse Signal 75 4.4.1 Reconstruction of the pulse signal 75 4.4.2 Filtering 76 4.5 Software 80 4.5.1 System overview 80 30 30 30 32 34 36 39 43 44 VI 4.5.2 LED control and system timing 81 4.5.3 Pulse arrival detection routine 83 4.5.4 Noise detection 91 4.5.5 Two-step averaging process 95 4.5.6 User interface and output 97 4.7 Sources of Error 98 4.8 Compliance with Electrical Safety Standards 99 5 LABORATORY EVALUATION OF THE PROTOTYPE SYSTEM 101 5.1 Calibration 101 5.1.1 Calibration method 101 5.1.2 Performance of the pulse sensor 104 5.1.3 Performance of pulse arrival detection algorithms 107 5.1.4 Performance of noise detection algorithms 109 5.1.5 Accuracy of PWTT values calculated by the system 111 5.2 Laboratory Testing 112 5.2.1 Purpose 112 5.2.2 Causing blood pressure shifts 113 5.2.3 Exercise-based stress testing 115 5.2.4 Results of exercise-based tests 116 6 INITIAL CLINICAL EVALUATION OF THE PROTOTYPE SYSTEM 120 6.1 Preparations for Clinical Trials 120 6.1.1 Obtaining approval for study 120 6.1.2 Test subjects 122 6.2 Experimental Protocol 123 6.3 Performance of the System in Obtaining PWTT Values 124 6.4 Results of Clinical Trials for Ascertaining the Validity of the Developed Adaptive Tourniquet Pressure Algorithm 125 6.4.1 Short cases on relatively young patients 125 6.4.2 Long cases under local anesthetic 130 6.4.3 Transient blood pressure changes 134 6.4.4 Other findings 134 vii 7 IMPROVEMENTS FOR APPLICATION AS A CLINICAL TOOL 136 7.1 Conclusions from Laboratory and Clinical Evaluations 136 7.2 Integration of a Standard Oscillometric BP monitor with the PWTT Measurement System 137 7.2.1 Description of the integrated system 137 7.2.2 Algorithm for adaptive tourniquet operation 139 7.2.3 Results of retrospective application of apparatus to data from clinical trials 142 7.2.4 Results of application of apparatus to new data from prospective cases 145 7.3 Considerations Regarding the Possible Use of the Pulse Signal from Pulse Oximeter for Pulse Arrival Detection 147 8 CONCLUSIONS AND RECOMMENDATIONS 151 8.1 Conclusions 151 8.2 Contributions of the Research 156 8.3 Recommendations and Topics for Further Research 157 REFERENCES 159 APPENDICES 164 Appendix A - Circuit Diagrams for Experimental System 164. Appendix B - Software Listings and Flow Charts 168 Appendix C - Chart Tracings For System Calibration 191 Appendix D- Results of Laboratory Testing 201 Appendix E - Approval Forms and Results from Clinical Testing 210 viii LIST OF FIGURES 2.1 Typical pressure distribution in tissue under a tourniquet. 1 4 3.1 The arterial system. 31 3 .2 Diagramatic representation of (A) pressure and (B) flow in the ascending aorta of an adolescent human subject. 3 2 3.3 Comparison of central and various peripheral arterial pulses from a normal male. 3 5 3.4 The heart chambers. 3 6 3.5 The ECG cardiac cycle. 3 8 3 .6 Errors associated with calculating slopes using two separated points. 4 5 3 .7 Modeling the arterial path as a series of discrete sections. 4 6 3 .8 PWTT measured between the R-wave and a pulse sensor located at a peripheral point on the body. 4 7 3.9 The effect of limb elevation on blood pressure and PWTT. 4 9 4.1 Block diagram of the prototype PWTT measurement system. 51 4 .2 R-wave detection circuitry. 5 4 4 . 3 The absorption of light in the vascular bed. 5 9 4 .4 Reflectance and transmittance type photoelectric sensors. 61 4 .5 Scattering and absorption coefficients of light in human tissue. 6 3 4 .6 The experimental photoplethysmographic pulse sensor. 6 7 4 . 7 Circuit for the LED current source. 6 9 4 .8 Circuit for biasing and sampling of the photodiode signals. 7 0 4 . 9 Gel housing for application of the arterial pulse sensor under proximal edge of a tourniquet cuff. 7 4 4 . 1 0 Frequency characteristics of input filter stages. 7 9 4 .11 Typical pre-filtered and post-filtered arterial pulse signals. 7 9 4 . 1 2 Block diagram of software routines. 8 2 4 . 1 3 Pulse timing functions. 8 3 4 . 1 4 Detection of leading edge of the pulse through slope upturn. 8 5 4 . 1 5 Potential error in the use of slope upturn detection. 8 6 4 . 1 6 Detection of the arrival of the leading edge by peak acceleration. 8 7 4 . 1 7 Arrival detection algorithm. 8 9 IX 4.18 The use of multi-sample differentiation to attenuate the second order response of small amplitude high frequency noise. 90 4.19 Pulse characteristics used for noise rejection routines. 93 4.20 Block diagram of two-step averaging process. 96 5.1 Sample trace from chart recorder as configured for calibration of the PWTT measurement system. 103 5.2 Sample of reversed polarity waveform. 105 5.3 Determination of pulse arrival detection point. 108 5.4 Performance of noise detection routines. 110 6.1 PWTT vs. BP - BD1. 126 6.2 PWTT vs. BP - BD2. 126 6.3 PWTT vs. BP - BD3. 126 6.4 PWTT vs. BP - BD4. 126 6.5 PWTT vs. BP-BD5. 127 6.6 PWTT vs. BP - BD7. 127 6.7 PWTT vs. BP-BD8. 127 6.8 PWTT vs. BP - BD9. 127 6.9 PWTT vs. BP-BD10. 128 6.10 PWTT vs. BP-BD11. 128 6.11 APWTT vs. ABP - combined cases. 129 6.12 Error in estimated ASBP as a function of APWTT. 130 6.13 PWTT vs. BP - RM2. 132 6.14 PWTT vs. BP - RM3. 132 7.1 Flow chart for operation of the proposed adaptive surgical tourniquet system controller. 140 7.2 Sample case for retrospective testing of the operation of the improved system. 143 X LIST OF TABLES 2.1 List of recommended tourniquet pressures. 15 3.1 Arterial pulse shape characteristics. 34 4.1 Main sources of error in measured PWTT. 98 6.1 Sample subjects for initial clinical testing. 122 7.1 Summary of results from retrospective study. 144 7.2 Sample subjects for prospective study of proposed apparatus. 145 7.3 Summary of results from prospective study. 146 xi MEDICAL TERMINOLOGY arterial blood pressure - the pressure exerted by the blood on the walls of the arteries, time varying with the cardiac cycle and the exact waveshape is a function of the point at which the measurement is taken. arthroplasty - an orthopedic surgical procedure where a joint is replaced with a prosthesis. arthroscopy - an orthopedic surgical procedure where the interior of a joint is viewed through a needle like camera which is inserted through a small hole in the skin. artifact - noise or corruption in a biological signal which originates from a source not associated with the observed signal source. BP - arterial blood pressure bpm - beats per minute, the measure of heart rate. diastolic blood pressure (DBP) - the term commonly used to refer to the minimum value of arterial blood pressure which occurs during diastole, the relaxation phase of the heart cycle. collagen - protein that occurs in the white fibers of connective tissues and in the matrix of bone. distal - further from the centre of the body, i.e. the hand is distal to the elbow. elastin - protein that comprises the yellow elastic fibers of connective tissue. epidural - the space between the dural sheath of the spinal cord and the bone of the vertebral canal exsanguination - removal of blood from within a portion of the body, usually by squeezing. extension - manipulation of a limb joint which straightens the limb such that the angle between the two bones is increased. femur - the thigh bone. XII flexion - manipulation of a limb joint which bends the limb such that the angle between the two bones is decreased. hypertensive - abnormal chronic elevation of arterial blood pressure. hypotensive - abnormal chronic depression of arterial blood pressure. induction - the act of inducing anesthesia through the administration of the appropriate anesthetic agents. intubation - introduction of a tube into the windpipe in order to carry the anesthetic/oxygen mixture to a patients lungs. invasive - any measurement or surgical procedure which requires penetrating the body. ischemia - deficiency in blood flow to a portion of the body. limb occlusion pressure - the minimum pressure of a tourniquet required to prevent blood flow under the cuff. mean arterial pressure (MBP or MAP) - the mean value of the arterial blood pressure waveform. noninvasive - in terms of a physiological measurement, any technique in which the monitored parameter is observed from outside the body. normotensive - normal, healthy blood pressure given the age and sex of the subject. oscillometry - a technique for estimating arterial blood pressure by observing the amplitude of small fluctuations in a pneumatic cuff applied to the upper arm. patella - the knee cap. peripheral resistance - the resistive component of the hydraulic load placed on the heart by the vascular system. proximal - closer to the centre of the body, i.e. the elbow is proximal to the hand. pulse pressure - the difference between systolic and diastolic blood pressure. radial artery -the superficial artery palpable at the wrist. XIII systolic blood pressure (SBP) - the term commonly used to refer to the maximum value of arterial blood pressure which occurs during systole, the contraction phase of the heart cycle. Valsalva maneuver - voluntarily increasing thoracic pressure by exhaling against a closed glottis. xiv ACKNOWLEDGMENTS I would like to thank Dr. James McEwen for his guidance and encouragement throughout the course of this project and for providing me with the unique opportunity to carry out this work at the Jack Bell Research Centre Facilities. I am indebted to Dr. Charles Laszlo opening the door to the fascinating field of clinical and medical engineering and for his guidance during the academic portion of my studies. I extend my deepest gratitude to Dr. Robert McGraw, Dr. Brian Day and Dr. Glenn Manning for their assistance and participation in the clinical testing and I am particularly thankful to Mike Jameson and Ken Glinz for their technical assistance and as constant sources of entertainment, advice and support over the past eighteen months. I am also indebted to the Vancouver Hospital and Health Sciences Centre and University of British Columbia Hospital O.R. staff for their assistance and tolerance during clinical trials. Finally, I must thank my friends and sometimes experimental subjects Jill Bryan, Jim Ross, Brian Bapty, Carol Harris-Jaeger, Steve Small, Dave Schmaltz and Mickey Thomas for their invaluable assistance, advice and encouragement. Financial support during this research was provided by IRAP through Western Clinical Engineering Ltd.. Chapter 1: Introduction 1 CHAPTER 1 INTRODUCTION 1.1 Motivation for the Research Surgical tourniquet systems are utilized to prevent blood flow into a limb during surgery so that the surgeon may have a bloodless field in which to operate. Surgical tourniquet systems include the following elements: a pneumatic tourniquet cuff for encircling the limb; a source of pressurized gas; and a controller connected to the gas source for controlling pneumatic pressure in the tourniquet cuff and thus indirectly the pressure applied by the cuff to the limb. Modern surgical tourniquet cuffs consist of an internal pneumatic bladder that is long enough to completely encircle the limb, and is enclosed in a woven, inextensible sheath. In addition to having a bladder which completely encircles the limb, a typical tourniquet cuff differs significantly from a blood pressure cuff in its design, shape and width. Typically the tourniquet cuff is applied above the surgical site and then inflated to stop blood flow into the limb past the cuff. Modem tourniquet controllers generally include a means for the operator to monitor and control the pneumatic pressure in the tourniquet cuff and the time of application of the tourniquet pressure. In spite of the almost universal use of surgical tourniquet systems in surgery of the extremities, their utilization has been accompanied by reports of limb paralysis, nerve damage and other injuries. It has been shown that there is some damage to the underlying tissues with each use of a surgical tourniquet system, and the extent of this damage increases as the level of pressure applied by the tourniquet cuff increases and as the time of pressure application increases. Optimally, it is generally accepted that in order to minimize the Chapter 1: Introduction 2 amount of injury to the limb, a tourniquet cuff should be inflated to the minimum pressure necessary to prevent blood flow into the limb during a surgical procedure and should be applied for as short a time period as possible. In practice, the optimum is not achieved and tourniquets are often inflated to arbitrary fixed pressures of approximately 250-300 mmHg for upper limbs and 300-500 mmHg for lower limbs. "Limb occlusion pressure" (LOP) is defined to be the inflation pressure required in a tourniquet cuff to stop or "occlude" arterial blood flow past the cuff to the distal portion of the limb. LOP is a function of a range of variables relating to the manner in which the pneumatic pressure in the tourniquet cuff results in the application of a pressure distribution to the surface of the underlying limb, as well as the manner in which this surface pressure distribution in turn produces pressures within the limb around the underlying arteries. The variables affecting LOP measurements are: tourniquet cuff location; tourniquet cuff design; width to circumference ratio of the cuff; fit of the tourniquet cuff relative to the shape of the limb; snugness of the cuff; tissue composition in the underlying limb; and blood pressure. LOP is therefore different from subject to subject and may be different for each application of a tourniquet cuff of a given design to the same subject. "Blood pressure" (BP) is defined to be the pressure of the blood within the arterial system and is measured in terms of two basic components: "systolic blood pressure" (SBP) corresponding to the peak pressure produced by the contraction of the heart; and "diastolic pressure" (DBP) corresponding to the pressure during the resting phase of the heart cycle. These values vary from point to point in the arterial tree and may be measured directly by inserting a pressure sensor into the artery. More commonly, blood pressure may be estimated indirectly by variants of three methods involving the use of a pneumatic BP cuff applied to the upper arm. Chapter 1: Introduction 3 These three methods all involve the inflation of the pneumatic BP cuff to collapse the arteries in the upper arm. It has been recognized that in order to yield accurate estimates of BP, the BP cuffs employed must conform to detailed design standards governing their design, construction, shape, bladder width, bladder length and other parameters. In present practice, LOP's typically are not measured prior to the inflation of a tourniquet cuffs for surgery. If the LOP's are measured using common tourniquet cuffs applied to a range of subjects, values of 120-180 mmHg for LOP at the upper arm and values of 150-300 mmHg for LOP at the thigh are typical. These LOP values are significantly lower than the tourniquet inflation pressures traditionally utilized in surgical applications. The higher traditional levels have been chosen to provide a large safety margin to ensure reliable occlusion in all tourniquet applications, especially if LOP is not measured prior to selecting the tourniquet inflation pressure, and to account for intraoperative increases in BP which will result in an increase in LOP. If the LOP is measured at the outset of a surgical procedure it can be expected that under normal circumstances, of all the variables which affect LOP, changes in BP will be the principal variable which will change the LOP over the duration of a surgical procedure. Therefore, by setting the tourniquet inflation pressure to the measured LOP plus a small safety offset, and by changing the tourniquet inflation pressure in response to measured changes in BP, potential exists to allow the use of tourniquet inflation pressures that may be on average much lower than the levels traditionally employed . If so, this should reduce the probability and extent of patient injury. This suggests the development of an "adaptive tourniquet system" consisting of a tourniquet and a controller that automatically adjusts the tourniquet inflation pressure based on estimated LOP, wherein the estimated LOP is Chapter 1: Introduction 4 calculated on the basis of a pre-operative measurement of LOP which is subsequently updated as a function of monitored changes in the patient's BP. In order for an adaptive tourniquet system to operate in a safe and effective manner it is necessary to be able to obtain frequent, accurate estimates of BP from the patient to allow detection of rapid shifts in BP that might otherwise cause blood to leak under the tourniquet cuff into the operating field. While an arterial catheter can provide continuous monitoring of BP, it is not suitable for general use because it is an invasive measurement which has its own associated hazards. BP measurements based on the use of a standard BP cuff do not provide continuous estimates of BP. Intermittent and frequent estimation of BP by repeatedly inflating and deflating a BP cuff may cause damage greater than that which would be expected to be avoided by reducing tourniquet pressure. In this thesis it is suggested that the use of measured "pulse wave transit time" (PWTT) may provide a means to estimate changes in BP to overcome the above problems and be suitable for control of tourniquet adaptation. PWTT is defined as the time required for the leading edge of the arterial pressure pulse to propagate between two points in the arterial tree. Studies have shown that changes in BP are linearly related to changes in the velocity of propagation along the arterial walls of the pressure wave resulting from each beat of the heart. Therefore, by measuring the change in the time it takes for the pressure pulse to travel between two fixed points in the arterial tree it may be possible to estimate changes in BP. This technique is ideally suited for use by an adaptive tourniquet system to estimate intraoperative changes in LOP since it allows continuous, non-invasive estimation of changes in BP and does not require the use of a BP cuff. In addition, potential exists for integrating information obtained from equipment commonly utilized in the operating room to allow the Chapter 1: Introduction 5 measurement of PWTT to the finger tip without applying additional sensors to the patient. The results of prior clinical testing of commercially available BP monitors which use PWTT to estimate BP suggest that such monitors may be able to produce BP estimates with errors equivalent to the other non-invasive techniques noted previously. 1.2 Scope of the Research The overall objective of the work in this paper was to investigate the use of measured PWTT to estimate LOP for use in the adaptive control of a surgical tourniquet. The specific objectives were as follows: 1) to investigate the physiological parameters relating to LOP and the minimum tourniquet pressure that can be safely utilized in surgeries; 2) to thoroughly investigate the theoretical basis for the estimation of blood pressure changes through measurement of pulse wave transit time and determine whether and how the technique can be adapted for use to estimate LOP; 3) to design, build and test a non-invasive arterial pulse sensor for producing an arterial pulse signal indicative of the arterial pulse sensed at a peripheral point on the body; 4) to develop a filter, a noise rejection algorithm and a detection algorithm for processing the arterial pulse signal in order to identify the arrival of the leading edge of the pressure pulse; 5) to design, build and test a microcontroller-based system that will provide continuous beat to beat measurement of PWTT at a point on the periphery utilizing signal inputs from a standard ECG monitor and the arterial pulse sensor as designed; and 6) to perform laboratory and clinical testing on a range of subjects utilizing the measurement system to analyze the relationship between changes in PWTT and Chapter 1: Introduction Q changes in BP and to employ this relationship in the development of an algorithm using measured PWTT to estimate both the LOP and the minimum tourniquet pressure required to safely stop blood flow in the limb. 1.3 Contributions of the Research The primary contributions of this research apply to the development of a safe and effective adaptive tourniquet system which will result in improved patient safety by permitting lower average tourniquet pressures to be employed. The most important contributions of this work are: 1) the development of a improved understanding as to the relation between PWTT and LOP; 2) the development and validation of a prototype PWTT measurement system suitable for use with an adaptive surgical tourniquet system; 3) the identification of limitations in the use of PWTT for the estimation of BP in bedside patient monitors or home monitoring devices which use preset coefficients in their estimation algorithms, or coefficients obtained through the determination of a two-point slope based on pre- and post-exercise BP and PWTT values; 4) the discovery of limitations in the use of PWTT based algorithms to accurately estimate BP in elderly patients or those undergoing drug therapy; 5) the development of a new algorithm for the continuous estimation of LOP and the minimum safe tourniquet inflation pressure based on the integration of continuously measured PWTT values with periodic BP estimates available from an oscillometric blood pressure monitor commonly available in operating rooms; 6) integration of this algorithm into an apparatus for operation as part of an adaptive tourniquet controller which estimates intraoperative LOP based on the measurement of Chapter 1: Introduction J PWTT using existing operating room equipment, which will thereby allow the use of lower and safer average tourniquet pressures; and 7) the initial evaluation and validation of the apparatus, both in the laboratory and the operating room. 1.4 Thesis Overview The following chapter provides background as to the physiological and practical aspects of the use of surgical tourniquets and the factors which affect LOP. The theory behind the use of an adaptive tourniquet is presented and a review of BP measurement techniques is presented. Chapter 3 is dedicated to the theoretical background for relating changes in PWTT to changes in BP and the development of an algorithm to estimate LOP and minimum safe tourniquet pressure from measured PWTT. The development of a prototype arterial pulse sensor and microcontroller-based system for measurement of PWTT is detailed in Chapter 4, and the results of preliminary laboratory testing to calibrate and evaluate the operation of the system is contained in Chapter 5. Chapter 6 details the protocol and procedures for clinical testing of the PWTT measurement system and contains analysis of the results obtained. A protocol for the use of PWTT measurement in conjunction with conventional periodic BP monitoring to develop subject specific linear regression models relating PWTT to SBP and LOP is presented and evaluated in Chapter 7. Chapter 1: Introduction 8 Chapter 8 reviews the conclusions of the research and the suitability of a system based on PWTT measurement for tourniquet control. The chapter also provides suggestions for commercial applications of the system as well as topics for further investigation. Chapter 2: Background and Review of Previous Research g CHAPTER 2 BACKGROUND AND REVIEW OF PREVIOUS RESEARCH 2.1 Surgical Tourniquets 2.1.1 Background The use of pneumatic tourniquets has become the standard means for providing a bloodless surgical field in a wide range of surgical procedures commonly performed on arms and legs. The use of a tourniquet to occlude blood flow while performing surgery on the limbs dates back to the time of the Romans [1], when tourniquets were used for limb amputations. The first pneumatic surgical tourniquet system was introduced by Harvey Cushing in 1904 and consisted of a rubber tourniquet cuff inflated using a modified bicycle pump which was applied to encircle the operative limb. A reservoir was integrated into the tourniquet system to help maintain the required pressure and a mercury manometer was utilized to monitor the air pressure within the tourniquet cuff [2]. This system replaced the use of an Esmarch bandage, a strong flat rubber bandage which was commonly utilized as a tourniquet by wrapping the bandage several times around the limb to prevent blood flow underneath. The use of an Esmarch bandage for this purpose holds significant risk of peripheral nerve injury due uncontrolled pressure being applied to the underlying tissues. This pressure can reach levels as high as 1000 mmHg at the midpoint of the bandage [3]. As a result, pneumatic tourniquet systems are now used routinely to better control applied pressures. 2.1.2 Modern pneumatic tourniquets Modern pneumatic tourniquets consist of an inflatable bladder of rubber or similar material of sufficient length to completely encircle the limb, covered in a woven inextensible sleeve. The Chapter 2: Background and Review of Previous Research -| q tourniquet is applied around the circumference the limb and held securely in place utilizing velcro or similar fasteners. A surgical tourniquet cuff is generally more rigid than a standard BP measurement cuff, often having plastic stiffeners within the sheath which provide improved pressure distribution to the tissues and durability for prolonged application. A tourniquet may be either cylindrical or tapered such that is assumes a conical shape more closely resembling the shape of the limb. Gas for inflation of the tourniquet is provided via one or two flexible hoses from a tourniquet controller which allows the surgical staff to control inflation and deflation of the cuff to the desired pressure as well as a providing visual display of the tourniquet bladder pressure. More elaborate microprocessor-based tourniquet controllers may provide timing features to record the duration of application and provide alarms in the case that preset maximum time limits have been exceeded or a cuff or hose failure is detected. 2.1.3 Method of application In preparation for a surgical procedure, the surgical tourniquet cuff is applied around the extremity and inflated to compress the underlying blood vessels in order to prevent the flow of blood to or from the limb distal to the point of application. This is to ensure that once the distal portion of the limb has been exsanguinated, the surgeon may perform the operation without the need to remove additional blood from the surgical field. The tourniquet cuff is generally applied snugly over several wraps of soft wide bandage and is isolated from the sterile operative field by a surgical drape. The cuff is applied as close to the surgical field as practical although care is taken to avoid application close to the knee or elbow joints where the sciatic and radial nerves respectively are vulnerable to compression injuries due to their close proximity to the skin surface and the underlying bone structure [1]. Exsanguination of the limb is performed prior to inflation of the tourniquet by either elevating the limb in order to Chapter 2: Background and Review of Previous Research <\<\ gravity "drain" the blood from the distal area [4] or through the use of an Esmarch bandage started at the distal end of the limb and wrapped spirally around the limb towards the trunk in order to "squeeze" the blood from the area. Often surgeons use a combination of both techniques. Once the limb area has been adequately exsanguinated, the tourniquet cuff is inflated and the limb lowered or the Esmarch bandage removed. The result is a bloodless surgical field which can be maintained for 1-3 hours before it is necessary to release the tourniquet to allow blood flow to the limb in order to minimize the risk of injury from ischemia. After blood flow has been restored for 10-20 minutes, the limb can then be re-exsanguinated and the tourniquet cuff re-inflated for surgical procedures which require longer times to complete [5-8]. There is no universally accepted maximum time for which a tourniquet can safely be left inflated. The tourniquet may also be utilized to facilitate the use of regional anesthesia injected into the limb without allowing the anesthetic to spread into general circulation. 2.1.4 Safety problems in the use of tourniquets Although the use of a tourniquet is almost universal in surgery of the limbs, tourniquet use has been accompanied by reports of limb paralysis, nerve damage and similar injuries. While the use of too low pressure in the cuff may cause some problems such as bleeding in the surgical field, passive congestion, edema, hemorrhagic infiltration of nerve tissue or premature release of regional anesthesia [5], the majority of cuff injuries are associated with overpressurization. Excessive pressure in the cuff or prolonged application can lead to post-operative paralysis of the limb, post-operative muscle weakness and pain at the cuff site, edema, and injuries of nerves, blood vessels, muscles and skin which appear to be the combined result of compression and local ischemia and may interfere with functional recovery Chapter 2: Background and Review of Previous Research 12 of the patient [1-21]. Soft tissue injuries such as bruising and blistering particularly at the distal edge of the cuff and post operative swelling may also result. These all appear to be a direct effect of applied pressure and unduly high pressure is thought to produce injuries more rapidly. It has been suggested that the true incidence of tourniquet induced complications may be significantly underestimated and that each use of a tourniquet may lead to some minor damage to the underlying tissue although in most cases the injury is short term and reversible and often masked by post-operative pain resulting from the trauma of the surgical procedure itself [11]. One study found that electromyographic (EMG) abnormalities occurred in approximately 70% of the lower limbs and 77% of the upper limbs of patients following surgery involving "normal" use of a surgical tourniquet compared to 0% and 3.4% respectively in procedures where a tourniquet was not utilized [2,20]. Complications resulting from tourniquet application may also be purposely underreported due to concerns of legal liability [12]. Since the incidence and severity of nerve injuries due to tourniquet use are a function of the applied pressure and time for which this pressure is applied [3], it is generally accepted that in order to minimize the risk of injury to the patient, the tourniquet cuff should be inflated to the minimum pressure required to ensure adequate occlusion and be applied for as short a duration as possible [1,4,16,18,20]. 2.2 Limb Occlusion Pressures 2.2.1 The selection of tourniquet pressure and factors which affect LOP The minimum inflation pressure of a tourniquet cuff to occlude arterial flow to the distal portion of the limb is difficult to predict since a tourniquet cuff does not apply uniform pressure to all points within the limb and no simple relationship exists between the inflation pressure of the Chapter 2: Background and Review of Previous Research -| 3 bladder and the pressure induced within the deep tissues and vessels [9]. LOP is a function of a range of variables relating to the manner in which the pneumatic pressure in the tourniquet cuff results in the application of a pressure distribution to the surface of the underlying limb and the manner in which this surface pressure distribution in turn produces pressures within the limb around the underlying arteries including: tourniquet cuff location; tourniquet cuff design; width-to-circumference ratio of the tourniquet cuff; fit of the tourniquet cuff to the shape of the limb; snugness of application; tissue composition of the underlying limb; position of the limb relative to the heart; and BP. Given that BP varies over the cardiac cycle, the pressure required to occlude blood flow then varies in time as a function of BP with the maximum value corresponding to the occurrence of the peak pressure point of the arterial blood pressure cycle which corresponds to SBP. LOP is therefore defined as the minimum tourniquet inflation pressure required to occlude blood flow at all points in the cardiac cycle. The transmission of pneumatic pressure within a tourniquet cuff to the underlying tissues has been described previously for normal limbs [3,16]. The pressure profile of the limb tissue is described by a parabolic curve, reaching its maximum value at the centre of the cuff and tapering off greatly at the edges with the mean soft tissue pressures consistently lower than the inflation pressure of the applied tourniquet [13]. The shape of the curve is affected by cuff design and by the shape match of the cuff to the limb. The differences between the soft tissue pressures at the middle and edges of the cuff increases with cuff pressure and high tourniquet pressures may lead to shear forces that can result in damage to the underlying nerves [13,16,18]. Most nerve injuries occur at the edge of the cuff where this pressure gradient occurs and tissue deformation is maximal [18]. DISTANCE TROU PROXIMAL EDGE OF CUFF (cm) DISTANCE FROU PROXIUAL EDGE OF CUFF (cm) TISSUE FLUID PRESSURE (mmHg) OISTANCC FROU PROXIUAL CDCE OF CUrF (cm) DISTANCE TROU PROXIUAL EDGE OF CUFF (cm) Distribution of tissue-fluid pressures in cadavers at four tissue depths beneath a 12-cm cuff on arms (left graph) and an 18-cm cuff on legs (right graph) applied at (A) 200 mmHg and (B) 400 mmHg Fig. 2.1: Typical pressure distribution in tissue under a tourniquet [18]. Recommended tourniquet pressures vary widely. In the relevant literature typical suggestions range from 225-300 mmHg for upper extremity surgery to as high as 500 mmHg for adult Chapter 2: Background and Review of Previous Research -j 5 lower limbs [2]. Table 2.1 details tourniquet pressure settings recommended in recent literature. Authors Pressure (mmHq) Time Limit (hr.) Cuff Width Bruner, 1951 270-300 1.0 none Boyes,1964 300 2.0 none Wilgis,1971 - 2.0 none Flat, 1972 250 2.0 none Love, 1978 - 2.0 none Milford, 1980 300 1.0-1.5 none Shaw, Murray, 1982 systolic + 100 - none Duthie, Bently, 1983 300 1.5 none Klenerman, 1980,1983 systolic +50 2.0-3.0 none Burton, 1983 225-275 - none Table 2.1: List of recommended tourniquet pressures [19]. The higher pressures are recommended due to smaller ratios of cuff widths to limb circumferences. While some adjustments are typically made to these recommended levels to take into account patient age, blood pressure, weight, and limb size, these recommended levels are significantly higher than normal peak LOP values determined through measurement; however, these high levels have been used traditionally to accommodate intersubject variability, intraoperative variation, hysteresis in pressure regulators, errors in gauges, leaks, poor cuffs and application techniques and intraoperative increases in BP which result in an increase in LOP in order to minimize the possibility of bleeding distal to the point of tourniquet application. Modern microprocessor-based tourniquet controllers have to a large extent removed the need for such large safety margins by allowing control of tourniquet pressures with a high degree of accuracy and reliability. Some of the recent literature Chapter 2: Background and Review of Previous Research 16 summarized in Table 2.1 suggests the use of patient SBP as a guide to setting the tourniquet pressure. The use of SBP plus a safety offset constant of 50 to 100 mmHg has been recommended [22,23]. These guidelines have been found to produce acceptable results in normotensive patients without grossly hypertrophied or obese limbs. Tourniquet pressures selected using guidelines based on measured SBP are generally lower than in the case that a "fixed" level is utilized. It is important to note that in recommending the use of pressure settings based on measured SBP few studies have discussed the effect of the design or width of the tourniquet cuff utilized These characteristics produce great variance in the results obtained using the BP-related techniques suggested above. Although LOP depends on BP, LOP is also a function of cuff design and cuff width relative to the size of the limb (conveniently described as the ratio of the cuff width to limb circumference or W/C ratio) with smaller values of this ratio producing marked increases in LOP. Studies have shown a relationship between this ratio and the percentage of cuff pressure transmitted to the underlying tissues [13,16,18]. If the cuff is too narrow, a low percentage of the bladder pressure is transmitted to the underlying tissues and a large pressure gradient is created across the tourniquet profile [19]. Wider tourniquets transmit a greater portion of the bladder pressure and result in a wider pressure "plateau" along the cuff profile [18,19]. It has been shown that on legs when the W/C ratio is less than 0.1:1, reliable occlusion could often not be obtained at pressures in excess of 400 mmHg, but a wider cuff with a ratio of 0.5:1 was found to occlude blood flow at pressures approaching SBP, and very wide cuffs with W/C ratios on the order of 1:1 were found to produce occlusion at pressures much lower than SBP [16]. Occlusion at pressures lower than SBP may be the result of the cumulative effect of flow resistance along a long length of narrowed artery. Narrow tourniquets may occasionally be used to allow bloodless access to Chapter 2: Background and Review of Previous Research -j j certain anatomical sites, but, it is generally accepted that the surgeon should choose the widest tourniquet which still provides an acceptable access to the surgical site. Cuff shape and fit has also been found to significantly affect LOP, with tapered cuffs producing occlusion at lower pressures than their cylindrical equivalents [20]. The difference in LOP is most likely a result of the fact that tapered cuffs may match limb shape better, particularly in the case of excessively muscular or obese individuals. This improved fit may result in increased contact area with the limb tissue which effectively results in a wider tourniquet cuff, i.e. a larger W/C ratio [20]. Studies in the laboratory have shown that for loosely or poorly applied cuffs and where the shape match of the cuff to the limb is poor, measured LOP is very high compared to a properly fitting tourniquet cuff of equivalent width. Therefore, choosing a tourniquet pressure based on a patient's pre-operative BP without taking into account tourniquet cuff width relative to limb circumference, tourniquet cuff shape, cuff fit and similar factors while generally better than the use of an arbitrary and fixed pressure setting will not result in of the optimum cuff pressure for individual cases. 2.2.2 The use of measured LOP for determining minimum effective tourniquet pressure A more accurate estimate of the minimum required tourniquet cuff pressure can be made by performing an actual LOP measurement on the patient prior to surgery. This measurement can be made by monitoring the pulse at a point distal to the point of tourniquet application using a Doppler flowmeter, photoplethysmographic pulse sensor or similar instrument and gradually inflating the tourniquet cuff until a pulse can no longer be detected at the distal point. The pressure at which the arterial pulse is stopped corresponds to the minimum Chapter 2: Background and Review of Previous Research 18 tourniquet cuff pressure required to occlude the underlying arteries or LOP at that time. The subject and limb should be resting horizontally as this measurement is taken. The use of a tourniquet pressure consisting of the measured LOP plus an additional offset to allow for intraoperative variables which may affect LOP will result in the use of a low and effective tourniquet pressure. It should also be noted that slightly different LOP values will be measured depending on whether LOP is determined as the point at which the arterial pulse ceases as the cuff is slowly inflated or if it is considered as the pressure at which the pulse resumes as the cuff is slowly deflated from a supra-occlusive pressure. This is because greater pressure is required to collapse the arterial vessels initially than is required to maintain them in this compressed state [10]. This difference is generally on the order of 3-5 mmHg. Reid et al. [2] found that the use of: Tpressurearm = LOP+50mmHg; and • pressure leg = LOP+75mmHg where LOP was measured using a Doppler pulse probe in conjunction with the surgical tourniquet after the patient had been anesthetized provided good homeostasis in 93.2% of upper limbs and 97.5% of lower limbs with no failures in 84 subjects. The resulting pressures utilized were an average of 190 mmHg for arms and 231 mmHg for lower limbs. The safety offset values utilized in this study were arrived upon subjectively after trial and error, and were required to account for intraoperative changes in patient BP and other Chapter 2: Background and Review of Previous Research 19 variables. Initial attempts to use a 50 mmHg offset in the lower limbs resulted in several failures which the author suggests may have been related to increases in patient SBP. Pedowitz et al [20] performed a similar study and found that using an offset of 50 mmHg in the lower limb resulted in only fair to poor homeostasis in approximately 20% of their test cases. Half of these fair to poor cases were believed to be related to increases in SBP after initial determination of LOP. While the advent of modern anesthetic techniques has reduced BP fluctuations in anesthetized patients, SBP can typically vary by 10-30 mmHg with larger shifts of up to 50 or 60 mmHg seen in certain cases. Once the surgical tourniquet is applied and inflated, all parameters significantly influencing LOP with the exception of BP can be expected to remain essentially constant for the duration of the surgical procedure. It can therefore be hypothesized that in the case that patient LOP is determined at the outset of an operation through the use of a Doppler flowmeter or similar means, any subsequent change in this LOP will be a direct result of a corresponding change in patient BP. The use of a tourniquet pressure P tourniquet may then be selected such that, where Ko is a small offset value to account for variables such as cuff movement relative to the ire (eq.2.1) limb and Kbi00d pressure is required to account for changes in BP. If it is possible to accurately monitor changes in BP over the course of the tourniquet application then Kb|00d ire can be modified adaptively in direct relation to changes in BP. Chapter 2: Background and Review of Previous Research 20 2.3 The Adaptive Tourniquet 2.3.1 The Development of an algorithm to allow the adaptation of a tourniquet based on measured LOP and the monitoring of intraoperative changes in SBP McEwen, who invented the microprocessor-based automatic tourniquet, also originated the concept of adaptive regulation of the tourniquet pressure with McGraw in 1982 [11]. Their opinion was that tourniquet pressure should be controlled as a function of time varying SBP. Given a minimum effective tourniquet pressure defined in equation 2.1 where both LOP and Koffset are fixed, calculating the ideal tourniquet pressure then becomes a matter of effectively determining the necessary change in tourniquet pressure to compensate for changes in SBP. Based on measured intraoperative BP changes the tourniquet inflation pressure can be adapted in one of two ways. Taking an arbitrary case involving a BP shift of +/- 20 mmHg these can be illustrated as follows. (1) Tourniquet pressure can be increased or decreased a percentage of the change in SBP from the initial value measured at the time of the direct LOP measurement. Given that depending on the physical properties of the tourniquet cuff, limb physiology and snugness of application of the tourniquet cuff, the pressure transmitted from the tourniquet cuff to the tissues surrounding the underlying arteries can range from less than 50% to almost 100%, the selection of the appropriate transfer function is difficult. The worst-case scenarios could be taken in order to ensure adequate occlusion is maintained such that for decreases in SBP, the pressure in the tourniquet cuff is reduced an equivalent amount, and for SBP increases pressure in the tourniquet cuff is increased by double the amount of the detected increase. For example, with initial LOP =200 mmHg: Chapter 2: Background and Review of Previous Research 21 case a/ SBP increases 20 mmHg, then increase PTOURNIQUET by 40 to 240 mmHg; case b/ SBP decreases 20 mmHg, then decrease PTOURNIQUET by 20 to 180 mmHg This approach, while theoretically ensuring adequate occlusion for a wide range of tourniquet applications, is not optimized to ensure that the minimum tourniquet pressure is used for each applications; or (2) Given the LOP as determined at the outset of the procedure, an estimate of the manner in which the tourniquet cuff pressure is transferred to the tissues surrounding the underlying arteries can be made by using an absolute SBP value determined at the same time as the LOP measurement to yield a LOP/SBP ratio which can then be utilized to allow estimation of a new LOP based on a change in SBP which relates to the specific physical parameters of each tourniquet application. Given that the pressure transmitted to the tissue surrounding the underlying arteries is approximately linear as a function of tourniquet pressure [13] then a given change in SBP will result in a change in LOP that can be estimated as: ALOP = ASBP * (LOP/SBP) where (LOP/SBP) is the characteristic ratio as described. Using this equation to estimate tourniquet pressures for the same +/- 20 mmHg BP change and an LOP measured as 200 mmHg with a corresponding SBP of 130 mmHg gives the following : PTOURN,QUET= LOP INITIAL + (LOPINMAL/ SBP,NMAL) * (A SBP) (eq. 2.2) Chapter 2: Background and Review of Previous Research case a/ PTOURNIQUET — 200+ 200/130* (20) = case b/ PTOURNIQUET = 200+200/130* (-20) = 230 mmHg 170 mmHg 22 In the above, inclusion of additional safety offsets (Ko) have been omitted for clarity. The second method allows the use of lower "optimized" tourniquet pressures by taking into account to the pressure distribution of the tourniquet as applied through the application of the characteristic LOP/SBP ratio as a multiplier. Equation 2.2 shall then be considered as providing a measure of the minimum effective tourniquet pressure, optimized for each individual tourniquet application. Equation 2.2 does however require the use of an initial absolute SBP value in addition to a knowledge of the relative BP change. It is interesting to note that an attempt was made to commercially market an adaptive tourniquet system similar to that described by McEwen and McGraw. The Richards Pressure Sentry ( Richards Medical Co., Memphis TN) employs both a limb occluding tourniquet and a conventional BP cuff. The pressure within the tourniquet is determined by adding an arbitrary selected offset pressure to SBP estimates obtained oscillometrically from the BP cuff which must be located on a nonoperative upper limb. The primary fault with this approach is that the Pressure Sentry is insensitive to the wide range of variables other than SBP which affect the LOP of the tourniquet. This two-cuff implementation can also be clinically inconvenient since the anesthetist also requires access to a nonoperative upper limb to apply a BP cuff for his own noninvasive monitoring equipment. These limitations in addition to other design limitations of the Pressure Sentry made it unpopular with surgeons and staff and it was subsequently withdrawn from the market [21]. Chapter 2: Background and Review of Previous Research 23 2.3.2 Clinical requirements of an adaptive surgical tourniquet system An adaptive tourniquet suitable for practical use must satisfy the following requirements: • Ideally blood pressure measurements should be made continuously or, if made periodically, updated at a frequency of no less than 4 times per minute in order to track fast moving increases in BP that might otherwise cause leakage under the cuff when using small safety offset values; • accuracy of BP measurements should be on the order of +/- 5 mmHg; • the system should be easily usable by non-technical personnel, with minimal user input required to apply, initiate and stop operation of the system; • any sensors required for attachment to the patient should be quick and easy to apply and relatively insensitive to positioning and such sensors should also be small and not intrude into the surgical field; • any sensors required for attachment to the patient should be relatively resistant to motion artifact; • all equipment must be impervious to liquids and resistant to being dropped, stepped on or otherwise roughly treated as is apt to occur in the operating room; • in the case that the BP measurement apparatus becomes unable to accurately determine BP, the tourniquet system must automatically revert to maintain a fixed and clinically acceptable pressure in the tourniquet cuff until accurate blood pressure estimations are once again available; and • operating personnel must be able to easily convert the system to fixed operation (i.e. by pushing a single button or switch). Chapter 2: Background and Review of Previous Research 24 Given continuous estimates of the minimum effective tourniquet pressure there are several possible modes of operation for an adaptive tourniquet. The system can be configured to continually track the LOP such that any perceived change in BP results in an immediate response from the tourniquet. The use of such continuous adaptation might be of annoyance to operating personnel due to the continual shifting of the limb as the tourniquet inflates and deflates. In addition, such continual adaptation of tourniquet pressure might in fact be more harmful to the underling tissues than would be the use of a fixed, higher pressure due to the shear effects of constant shifting against bone and compressed tissue; however, no studies to date have been done concerning this potential risk. As an alternative to continuous adaptation, it may be more suitable to set minimum thresholds such that the system adapts only in steps of 10 or 15 mmHg. The exact method of adaptation is not the subject of this thesis and shall be left for further study by others. 2.4 Blood Pressure Measurement Techniques 2.4.1 The physiology of blood pressure Contraction of the left ventricle of the heart (systole) to expel blood into an elastic arterial system creates a pressure wave in the systemic arterial system. The peak of this pressure wave corresponds to systolic blood pressure (SBP). This is the peak interior pressure seen in the underlying arteries at the proximal edge of a tourniquet. While the ventricle is relaxed (diastole) the pressure falls as the volume of blood expelled by the ventricle drains from the arterial system into peripheral circulation. The lowest point in this pressure cycle corresponds to diastolic blood pressure (DBP). In a resting young adult, typical pressures are 120 mmHg (16 kPa) systolic and 90 mmHg (12 kPa) diastolic, generally expressed as 120/90. These figures normally increase with age, particularly the systolic pressure as the arterial system becomes stiffer. BP also increases due to a variety of short term physiological changes such Chapter 2: Background and Review of Previous Research 25 as exercise or stress which can lead to more vigorous left ventricular contraction and an increase in heart rate [24,25]. BP varies slightly over the body in relation to the distance from the heart at which the measurement is taken. Studies using cannulas tipped with pressure sensors have found that the SBP within the radial artery at the wrist is generally on the order of 10% greater than the pressure produced by the same pulse at the aorta [26,27]. Conversely, diastolic pressures tend to be slightly less at the periphery. This results in a gradual increase in pulse pressure (the difference between systolic and diastolic pressure ) toward the periphery with a pulse pressure on the order of 145% found at the radial artery of the wrist as compared to the aorta. This increase in pulse pressure is a result of reflective phenomenon and increasing peripheral resistance discussed further in the next chapter. This amplification is highest among young subjects and decreases with age. In addition, the shape of the arterial pressure wave resulting form each contraction of the heart becomes more rounded with a marked decrease in rise time as it moves outward due to damping of higher harmonics by elastic arterial tissues [27-30] again this shall be discussed further in the next chapter. Low frequency cyclic variations have been found in blood pressure (10-20 second cycles) which do not correlate with respiratory or movement cycles. These may be related to vasomotor changes involved in thermal regulation as they show a marked increase in amplitude during exercise and may represent the body hunting for a balance between blood distribution to the skin for cooling and to the muscles for oxygen supply [26]. A patient undergoing a surgical procedure is positioned horizontally on the operating table such that all points along the main arterial paths to the arms and legs are at essentially equal elevation. The only limb to be elevated or have its position changed over the course of the Chapter 2: Background and Review of Previous Research 26 surgical procedure is the operative limb, which is occluded of blood and fully exsanguinated of all blood. Given that the physiology of the subject remains otherwise unchanged, any increase or decrease in BP will be seen proportionately at all points in the arterial tree, although the absolute pressure values show some difference from point to point. Therefore, measuring the percentage shift in BP at any point on the body is in essence the same percentage shift as would be seen at any other point. If the operative limb is elevated, the only resulting pressure changes in the general circulatory system will be a small increase in the overall system pressure caused by the small standing column of blood between the heart level and the proximal edge of the cuff. The body reacts very quickly to compensate for this increase in order to maintain the desired pressure levels in the arteries feeding the brain and vital organs. While intuitively one would think that elevating the legs would result in an increase in BP in the horizontal portion of the body, lab experiments have found that BP in the torso actually drops slightly as the body overcompensates for the sudden pressure increase. The BP measured at a point on.the elevated limb is less than the systemic blood pressure by a value that is roughly equivalent to the per unit mass of blood multiplied by the height of the standing column of blood between the measurement point and the heart. This has been found to be approximately 1 mmHg per cm of elevation. 2.4.2 Current technologies for the estimation of blood pressure Actual BP can only be measured directly by utilizing a pressure-sensing cannula which is inserted into the artery through an incision in the skin. This type of invasive measurement is not considered clinically acceptable for general use as it introduces hazards for the patient. More commonly, blood pressure may be estimated indirectly by variants of three methods Chapter 2: Background and Review of Previous Research 27 involving the use of a pneumatic BP cuff applied to the upper arm: oscillometry; auscultatory measurement using K-sounds; and flow measurement using a Doppler flow meter. These methods involve the inflation of the pneumatic BP cuff to collapse the arteries in the upper arm and therefore in order to yield accurate estimates the BP cuffs employed must satisfy recognized standards governing their design, construction, shape, bladder width, bladder length and other parameters. Of these methods of BP estimation, few are mechanically robust enough to withstand the rigors of an orthopedic procedure, sensitive enough for use with pediatric or obese patients and sufficiently insensitive to noise resulting from motion of the limb(s) during the surgical procedure. More importantly, none are suited to obtaining the frequent BP readings required to allow adaptive control of a tourniquet. Auscultatory approaches where a piezoelectric or similar contact microphone is utilized to detect the Korotkoff sounds associated with systolic and diastolic pressures are position sensitive, particularly sensitive to noise from motion artifact and require the use of either a special dual bladder tourniquet cuff or the use of a BP cuff on a nonoperative limb. Similar problems are found with the use of the Doppler ultrasonic flow technique as it is particularly important that the ultrasonic probe be located directly over the artery in order to detect the motion of the arterial wall due to pulsatile blood flow and the use of a second cuff is again required. The use of oscillometry is the standard means for monitoring BP in the operating room, relying on analysis of the oscillations in the pressure of a BP cuff applied to the upper arm induced by pulsations in the underlying arteries. Oscillometry BP monitors are typically applied to all subjects undergoing surgery and set to take readings every 2-5 minutes requiring approximately 30 seconds to obtain each estimate. Very small oscillations are noted with each Chapter 2: Background and Review of Previous Research 28 heart beat when the cuff is above systolic blood pressure. These suddenly increase in amplitude as the cuff pressure falls through the systolic point and reach a maximum when cuff pressure is at mean arterial pressure (MBP). Various algorithms are utilized to estimate DBP and SBP from the pulse amplitudes. SBP may be estimated as the point at which a large increase in pulse amplitude is observed or alternately as the point at which the pulse amplitude is approximately one half the maximum observed amplitude. Similarly the DBP is generally estimated as the point at which the pulse amplitude falls to 80% of the maximum observed amplitude [21]. The DBP and SBP values obtained through oscillometry are estimates based on MBP. Since the technique relies on obtaining a noise free train of pulses as the cuff pressure is decreased from supra-systolic to sub-diastolic pressures any motion of the limb will result in corrupted pulse stream. In addition the time required for each new measurement limits the suitability for adaptive tourniquet control. It is suggested that the use of measured "pulse wave transit time" (PWTT) may provide a means to estimate changes in BP to overcome problems with the above techniques and be suitable for control of tourniquet adaptation. PWTT is defined as the time required for the leading edge of the arterial pressure pulse to propagate between two points in the arterial tree. Studies have shown that changes in BP are linearly related to changes in the velocity of propagation along the arterial walls of the pressure wave resulting from each beat of the heart. Therefore, by measuring the change in the time it takes for the pressure pulse to travel between two fixed points in the arterial tree it may be possible to estimate changes in BP. This technique is ideally suited for use by an adaptive tourniquet system to estimate intraoperative changes in LOP since it allows continuous, non-invasive estimation of changes in BP and does not require the use of a BP cuff. In addition, potential exists for integrating information obtained from equipment commonly utilized in the operating room to allow the Chapter 2: Background and Review of Previous Research 29 measurement of PWTT to the finger tip without applying additional sensors to the patient. The results of prior clinical testing of commercially available BP monitors which use PWTT to estimate BP suggest that such monitors may be able to produce BP estimates with errors equivalent to the other non-invasive techniques noted previously [32-34]. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 3fj Transit Times to Estimate Required Tourniquet Pressure CHAPTER 3 THE ADAPTATION AND EVALUATION OF ALGORITHMS EMPLOYING PULSE WAVE TRANSIT TIMES TO ESTIMATE REQUIRED TOURNIQUET PRESSURE 3.1 The Arterial System 3.1.1 Physiology The arterial system is a complex network of branching tubes of varying elasticity. The structure and elasticity of the arteries gradually changes between the heart and the periphery. The thoracic aorta is an "elastic" artery. Its mechanical properties are determined by the multiple elastic lamellae contained within its medial coat. Smooth muscle is present in small amounts as is collagen but neither appears in sufficient quantity to have any roll in the mechanical properties of the arterial walls [28]. The abdominal aorta contains relatively more collagen in its media but is still regarded as an elastic artery. The major arteries arising from the aorta contain more smooth muscle and less elastin. Peripheral arteries such as the brachial, radial, femoral or tibial are predominantly muscular with little elastin in the wall while tiny peripheral arteries are almost entirely devoid of elastin [28,29]. With aging, this arrangement is progressively distorted as a consequence of thinning, splitting, fragmentation and fraying of the elastic fibers with collagen the primary element in remodeling the vessel wall and the result is a gradual stiffening of the arterial walls over time [28,29,30,35,36]. This is most evident in the aorta and thoracic arteries which are the most elastic in young subjects. This stiffening is such that the systemic arterial tree appears to Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave Transit Times to Estimate Required Tourniquet Pressure 31 Popliteal a Posterior tibial a Anterior tibial a. Carotid sinus Rt. vertebral a. Rt. subclavian a. Axillary a. Internal thoracic a. Brachial a Descending aorta Ulnar a. Radial a. Common iliac a Internal iliac a External iliac a Testicular (gonadal) a Deep femoral a Femoral a External carotid a. Internal carotid a. Lt. common carotid a. Lt. subclavian a. Brachiocephalic a. Arch of the aorta Coronary aa. Celiac trunk Intercostal a. Superior mesenteric a. Inferior mesenteric a. Radial a. Dorsal pedis a. Margu l ies /Wa ld rop Fig. 3.1: The arterial system [25]. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 32 Transit Times to Estimate Required Tourniquet Pressure contract physically, the greatest affect being seen in the ascending aorta which is foreshortened to the greatest degree [29]. Arterial radius and wall thickness increase with age until about the eighth decade and then remain relatively constant. Each contraction of the heart produces a pressure pulse which travels outward over the arterial walls to the periphery. The velocity of this pulse ranges from about 4 m/sec in the aorta to 12 m/sec in the radial and tibial arteries supplying the hand and foot respectively and is the sum of the absolute wave velocity and the velocity of blood flow within the arteries. Blood velocity is on the order of 2% of PWV [35] and varies significantly over the cardiac cycle. A digaramatic representation of pressure and flow in the ascending aorta is given in Fig. 2.2. (A) Fig. 2.2: Diagramatic representation of (A) pressure and (B) flow in the ascending aorta of an adolescent human subject [29]. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 3 3 Transit Times to Estimate Required Tourniquet Pressure 3.1.2 Wave reflections While ascending aortic pressure waveforms show only one spurt of flow from the heart during systole, the arterial pressure wave recorded in the ascending aorta or the peripheral arteries shows evidence of two and sometimes three separated waves. This is a result of wave reflection [29]. A pressure wave traveling along a vessel will be reflected at any discontinuity which causes a mismatch of impedance. Such potential sites in the arterial tree include: parts of the arterial wall where stiffness increases; branching points; or the periphery of the arterial tree where the low resistance arteries terminate in high resistance arterioles. Evidence of wave reflection can be virtually eliminated by vasodilation of the peripheral vascular bed through the injection of a vasodilator agent which suggests that of these sites, the main source of wave reflections is the artery-arteriole junctions [29]. Under normal conditions, this discontinuity results in approximately 80% of the incident wave being reflected [29]. With vasodilation, this reflection coefficient may approach zero but with intense constriction approaches unity. While terminal reflection of the pulse pressure wave may be on the order of 80%, one rarely sees evidence that the reflected primary wave approaches this proportion of the incident wave, even under conditions of intense vasoconstriction. This is because the reflected wave is attenuated in travel along the arterial tree and will be reflected once again upon reaching the upper body. A further factor in the distortion of this reflected wave is dispersion of the reflected waves due to the reflecting sites being located at different distances from the heart. Interaction of the waves that are reflected from these different sites will tend to broaden and lower the amplitude of the overall reflected wave. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 3 4 Transit Times to Estimate Required Tourniquet Pressure 3.1.3 Shape of the pressure pulse The shape of the arterial pressure pulse differs significantly depending on the location of the measurement point. Table 3.1, describes the variance of the average buildup time of the pulse, defined as the time from onset to maximal pressure, and the average width of the pulse defined as the time from the onset to the occurrence of the dicrotic notch (the small notch immediately following the peak magnitude as seen in Fig. 3.2) at various points in a population of healthy males. CENTRAL FEMORAL BRACHIAL RADIAL BUILDUP (msec) 166 138 109 102 WIDTH (msec) 311 336 320 330 Table 3.1: Arterial pulse shape characteristics [27]. It is evident that towards the periphery the velocity of the pulse wave increases and the buildup time is shortened. However the overall width of the pulse increases as it moves to the periphery. This can be better seen in the following traces of pulse waveforms taken at various locations. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave Transit Times to Estimate Required Tourniquet Pressure 35 RIGHT RADIAL LEFT RADIAL LEFT BRACHIAL AORTIC RIGHT FEMORAL RIGHT DORSALIS PEDIS ECG Fig. 3.3: Comparison of central and various peripheral arterial pulses from a normal male [27]. A striking difference occurs between the contour of the pulse wave in the brachial-radial and femoral-dorsalis pedis systems as seen in Fig. 3.3. A double systolic wave develops in the pressure pulse during transmission peripherally from the subclavian artery which results in an exaggerated dicrotic dip during transmission down the arm. This can be seen clearly in the brachial and radial arterial pulse waveforms in Fig. 3.3. In contrast, the pressure pulse at the femoral artery consists of a single systolic wave followed by a poorly defined or absent dicrotic dip [26,27,35]. The contour of the pulse wave measured at the same location on different individuals will also show significant variation. It is evident that there is significant variation in the pulse shape at different points in the arterial system. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 35 Transit Times to Estimate Required Tourniquet Pressure 3.2 The Cardiac Cycle and Interpretation of the ECG The heart serves as a four chamber pump for the circulatory system. The main pumping action is provided from the ventricles while the atria are essentially antechambers which store the blood during the contraction of the ventricles. Superior vena cava Right pulmonary veins Pulmonary semilunar valve Right atrium Tricuspid valve Chordae tendineae Right ventricle Inferior vena cava artery veins valve septum Fig. 3.4: The heart chambers [25]. The smooth rhythmic contraction of the heart is the result of a coordinated series of electrical events in the heart muscle tissue [25,37,38]. A set of electrodes placed on the surface of the chest area can be used to record the potential differences across the heart in order to provide a "picture" of the electrical events which indicate the activities of the heart. This recording, an Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 3 7 Transit Times to Estimate Required Tourniquet Pressure "electrocardiogram" (ECG or EKG ) consists of three distinct waves, designated P, QRS and T which represent specific parts of the cardiac cycle. These waves are not action potentials but represent changes in the potential between two regions of the heart which are produced by the composite effects of action potentials in many myocardial (muscle) cells. Atrial depolarization, and the associated filling of the atrium with blood is associated with the P wave. The following conduction of this impulse into the ventricles to constrict and expel the blood into the circulation system similarly creates a potential difference that results in a sharp upward deflection of the ECG waveform which returns to baseline as the entire mass of the ventricles becomes depolarized. This portion of the cycle is referred to as the QRS wave and signifies the ejection phase of the cardiac cycle. During this time, the atria repolarize but this is hidden due to the greater depolarization occurring in the ventricles. This is followed by the T wave, occurring due to the repolarization of the ventricles which signifies the beginning of diastole [24,25]. Given that the QRS wave indicates the onset of the ejection phase of the cardiac cycle it is this portion of the wave form which is best utilized in timing of pulse wave transit time. The peak of the wave ( "R") occurs at the point at which approximately half of the ventricles are depolarized, since the equal mass of polarized and unpolarized tissue leads to the maximum potential difference. The "S" point occurs at the point where the entire mass of the ventricles are depolarized and in contraction [25]. The delay between the onset of these potential differences and the actual ejection phase of the heart is on the order of 40 msec. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave Transit Times to Estimate Required Tourniquet Pressure 38 0 0.5 1.0 Time, s Fig. 3.5: The ECG cardiac cycle [37]. Since ECG signals result from action potentials produced in muscle contraction, they are vulnerable to noise resulting from movement of the arms , upper body or any other type of motion which involves flexing of the muscles of the chest region. In patients undergoing surgery such noise tends to be relatively small in amplitude since they are lying relatively still on the operating room table and is not expected to seriously effect the ability to identify the occurrence of the QRS complex. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 3 9 Transit Times to Estimate Required Tourniquet Pressure 3.3 The Theoretical Background for Relating PWTT to BP The velocity with which a pressure wave travels along an artery is determined by the mechanical properties of that artery and the relationship between the two has been described by equations derived from bench experiments in which an excised artery is utilized or modelled using thin walled elastic tubing [41]. The analysis of propagation of a pulse wave along the arterial walls is closely related to the analysis of the velocity of sound (c0) in air which was first studied by Newton who derived the equation, where K is the volume elasticity and p is the air density. The analogy between the velocity of wave propagation in a compressible fluid (i.e. air) and the velocity of a wave traveling in an incompressible fluid in an elastic tube has been proposed and experimentally validated [39,71] and allows the derivation of an equation for the velocity of the arterial pulse wave. When a body of length Xo is extended to a length x^ the relative increase in length (xi - Xo )/xo is the longitudinal strain. Strain in the y and z directions may be defined similarly. Stress is the intensity of a force acting on a given point in a body and may also be resolved into x,y,z components. Hooke's law states that, within certain limits, strain is proportional to stress [71]. The bulk modulus, B, is defined as the ratio of compressive stress to strain or equivalently the ratio of compressive stress to change in volume. Taking a cube under compression, the mean stress can be represented as a pressure, p, which will exert the same force in all directions (Pascal's law). Using the mean compressive stress p, the bulk modulus B may then be written as, c 0 =-lYJp (eq. 3.1) B = Ap * V / AV (eq.3.2) Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 40 Transit Times to Estimate Required Tourniquet Pressure where V is the initial volume and AV the change in volume. Substituting this equation for bulk modulus into Newton's equation in place of the volume elasticity of air K the following equation is obtained for the velocity of propagation of a wave in an elastic vessel: c = V (V* ( Ap/(AV» )) or c2 = (Ap/AV) * (V/p) (eq. 3.3) where c = PWV, Ap = change in the pressure within the artery, AV= volume change, V = initial volume, and p = density of fluid. This is the form of the equation derived and validated by Bramwell and Hill in 1922 [39], however their method of derivation is slightly different. Using the Moens-Korteweg equation for the transmission of a pulse wave derived from Newton's equation using the law of Laplace, c = V(( E*h)/(2*RV)J (eq.3.4) where E is Young's modulus in the circumferental direction, h is the wall thickness, and R is the wall radius, Bramwell and Hill applied a simple transformation to obtain eq. 3.3 as follows. A small rise Ap in pressure may be shown to cause a small increase, AR = R2Ap/Eh, in the radius R of the artery, or a small increase AV = 2nR3Ap/Eh, in the volume V per unit length. Hence (2R)/(Eh) = AV/(VAp), which when substituted into the Moens-Korteweg equation yields eq. 3.3. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 4 -| Transit Times to Estimate Required Tourniquet Pressure It can be observed from eq. 3.3 that PWV is inversely related to the square root of compliance or (Av/Ap) as well as the dimensions of the vessel. While it may be assumed that in the short term the initial volume of the vessel and the fluid density remains essentially constant in human arteries, compliance does not remain stable. PWV can therefore be used as an index of compliance which varies primarily as a function of [40]: age - as a person ages, a gradual stiffening of the arteries occurs decreasing compliance and increasing PWV; pathological condition - various diseases affect the compliance of the arterial walls such as diabetes or arteriosclerosis; drugs - various drugs such as beta blockers and vasodilators affect arterial compliance; and pressure - as the fluid pressure within the artery increases, the artery is stretched and stiffened, decreasing compliance and conversely the artery becomes more compliant as this pressure decreases. Given that age, pathological condition and to some extent medication levels can be expected to remain relatively constant over the duration of a surgical procedure, changes in compliance will then be caused primarily by changes in the blood pressure within the arteries. The fact that arteries and arterial segments become progressively more resistant to stretch (less distensible) at higher distending pressures has been recognized for over a century and subsequently confirmed in many studies. These studies have identified a near linear relationship between PWV and BP over a wide range of pressures [39,40,42,43] and this relation has also been found valid over a wide range of pulse frequencies [41]. The relation between PWV and blood pressure can therefore be described in terms of a slope and intercept as [32,33,34,43], Chapters: The Adaptation and Evaluation of Algorithms Employing Pulse Wave Transit Times to Estimate Required Tourniquet Pressure BP =d+ PWV * C2 where C? and C 2 are the intercept and slope coefficients respectively. PWV is a relative measure of BP, meaning that an initial calibration is necessary to determine and C 2 and thus estimate absolute pressure values. Variations in individual physiology require that this calibration be performed on each individual or that preset and mean values of Ci and C 2 be utilized. Given that the transit time of the pulse wave over a fixed arterial path is linearly related to PWV as a function of path length, equation 3.5 can be written in terms of PWTT as, BP = d + PWTT * C 3 (eq. 3.6) where the constant C 3 includes a factor related to path length and PWTT is the pulse transit time, generally on the order of a few hundred milliseconds depending on path length. Since it is only necessary to determine changes in blood pressure from a baseline level to control an adaptive tourniquet, the constant term can be ignored such that, ABP = A PWTT * C (eq. 3.7) where C is the slope coefficient for the given path. There is no general consensus as to whether better correlation to PWV is found with SBP, MBP or DBP. Some studies have found that DBP is better correlated [44-46] , while in other cases better results have been observed in correlation with systolic pressures [33,34,40]. 42 (eq. 3.5) Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 4 3 Transit Times to Estimate Required Tourniquet Pressure 3.4 The Development of an Algorithm for Estimating Ideal Tourniquet Pressure Based on Measured PWTT Given equation 3.7 relating changes in BP to measured PWTT and an LOP/BP ratio measured at the outset of a procedure the required change tourniquet pressure to compensate for a change in SBP can then be determined in terms of equations 2.2 and 3.7 as: APTOURNIQUET = (LOPINITIAL/ BPINITIAL ) * ( A BP) = (LOP,N| T IAL/ BP1N,TIAL ) * APWTT * C (eq. 3.8) Given that SBP, the peak blood pressure during the cardiac cycle, is more conveniently measured than either DBP or MBP, it was decided to use the measurement of SBP for testing of eq. 3.8. Given a safety offset POFFSET chosen as a buffer for fast changes in SBP and to account for other minor variables, then the minimum effective tourniquet pressure at any time during the course of the surgical procedure can be estimated in terms of measured PWTT as, PTOURNIQUET = LOPINITIAL + POFFSET + (LOPINITIAL/ SBP,N|TIAL ) * A P W T T * C (eq. 3.9) where C is the slope coefficient of the linear regression model relating PWTT to SBP over the chosen arterial path and APWTT is the change in transit time over the given path from the baseline value determined at the time of the LOP measurement. This equation may allow the estimation of the optimum tourniquet inflation pressure at any point during a surgical procedure. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 4 4 Transit Times to Estimate Required Tourniquet Pressure 3.5 The Selection of Slope Coefficients Relating APWTT to ASBP and the Selection of the Pulse Path for Measurement In modeling the PVW/SBP relation in terms of equation 3.9, the determination of the slope coefficient C can be made in either of two ways: 1. a generic slope coefficient can be utilized, determined based on studies on a sample population; or 2. slope coefficients that are specific to each individual subject ("patient-specific slope coefficients") can be utilized. In existing devices which employ measured PWTT to estimate BP these are determined by taking BP/PWTT measurements at two separate blood pressure levels (usually pre-exercise and post-exercise) and determining the exact slope coefficient value from these two points. A system for utilization in the operating room may be limited to the use of generic slope coefficients since obtaining two adequately separated BP values may not be possible. In addition, the use of a generic slope coefficient may actually produce more accurate results than in the case that patient-specific slope coefficients are determined using two separated points. One study found correlations between PWTT based estimates and BP estimates made through the use of a BP cuff and stethoscope to listen for K-sounds of 0.92 (SBP) and 0.97 (DBP) using a generic slope coefficient compared to 0.90 and 0.80 using patient-specific slope coefficients [33,34]. Although intuitively, this result may seem incorrect it makes sense in consideration of the effect of small errors in the measurements associated with determining patient-specific slopes. Given a possible error of +/- 3 mmHg in the BP measurement and +/-5 msec in PWTT measurement and using a true slope coefficient of 0.6 (shown to produce good results for relating R-wave onset - finger tip pulse arrival transit time to SBP), Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 4 5 Transit Times to Estimate Required Tourniquet Pressure calculations for a subject with for a 20 mmHg shift could yield slope coefficient values ranging between 0.32 - 1.13 as shown in Fig. 3.6. Fig. 3.6: Errors associated with calculating slopes using two separated points. Obviously in order to minimize this error it is advantageous to have as large a spread as possible in the separated BP values used for determining the slope. Since generic slopes are essentially different from the actual slope of the PWTT/BP function, calculation errors resulting from their use will increase in proportion to the change in a subject's BP. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 46 Transit Times to Estimate Required Tourniquet Pressure Since these slope coefficients show a linear dependence on the arterial path length, estimation of BP by measuring transit times to different locations such as the thigh, upper arm, finger or toe requires specific coefficients calibrated for each pulse sensor location and is the subject of clinical evaluation described in later chapters. Initially it might also appear that the effect of gradually narrowing arteries and branches of reduced size along the path of propagation might require the use of several a slope coefficients to model the PWTT/BP relation over the branches which make up the overall arterial path. However, looking at the path as a series of discrete sections, each with its own characteristic linear regression model coefficients, it is seen that the use of a single overall slope coefficient is valid over the combined path and can be determined easily in terms of overall PWTT as seen in Fig. 3.7. ARTERIAL S E G M E N T 1 ARTERIAL S E G M E N T 2 ARTERIAL S E G M E N T 3 1 APWTT, APWTTz APWTT 3 ABP = C1 * APWTT, APWTT, = ABP C, ABP = C 2 * A P W T T 2 APWTT 2 = ABP C2 ABP = C 3 * A P W T T 3 A PWTT3 = A B P C 3 APWTT-TOTAL = ABP + ABP + ABP ABP = C * APWTT TOTAL where C = Fig. 3.7: Modeling the arterial path as a series of discrete sections. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 4 7 Transit Times to Estimate Required Tourniquet Pressure In determining transit time, the lag between the R-wave and the actual pulse ejection (generally on the order of 40 msec and relatively insensitive to heart rate [40] ) should be noted. Since this value will remain relatively constant in a patient over the duration of a surgical procedure it will therefore not affect estimations of relative pressure changes as in equation 3.9 which doesn't take into account a constant BP term. In addition, the use of the longer path length through R-wave based timing reduces the significance of measurement errors associated with the detection of the of the pulse wave arrival compared to the use of two pulse sensors separated along the arterial path. The use of a single pulse sensor in conjunction with the signal from an ECG was therefore selected for development of a prototype measurement system to investigate the use of measured PWTT to estimate changes in LOP to control an adaptive tourniquet. R-WAVE ECG SIGNAL t ARTERIAL PULSE ARRIVAL AT PERIPHERAL POINT ARTERIAL PULSE SIGNAL PWTT TIME t = 0 t — Tarrival Fig. 3.8: PWTT measurement between the R-wave and a pulse sensor located at a peripheral point on the body. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave 48 Transit Times to Estimate Required Tourniquet Pressure The arterial pulse sensor may be located at a variety of locations including under the proximal edge of a the tourniquet cuff. While it may be expected that the use of a pulse sensor on a non-operative finger or toe would be capable of producing a strong arterial pulse signals and may allow estimation of BP changes without requiring the use of a wide range of slope coefficients relating to different arterial paths, the use of a pulse sensor in the cuff has the potential advantage of allowing the sensor and the cuff to be a combined device, removing the need for an "additional" sensor to be applied to the patient. In addition, locating the pulse sensor at the tourniquet cuff has additional potential to detect local pressure changes seen at the proximal edge of the cuff due to limb elevation or lowering by the surgeon. This additional potential is however limited because the change in PWTT observed for a given pressure change due to a change in limb elevation will be smaller than that that would be observed due to a comparable change in the overall systemic blood pressure. This is due to the fact that the pressure change will Only occur in the limb portion of the arterial path rather than over the entire path as it would in the case of a systemic BP change for a subject whose limbs remain level. See Fig. 3.9. Chapter 3: The Adaptation and Evaluation of Algorithms Employing Pulse Wave Transit Times to Estimate Required Tourniquet Pressure systemic SBP = 120 mmHg HEART/ systemic S B P = 90 mmHg HEART, systemic S B P = 120 mmHg HEART, TORSO THIGH PWTT = 300 msec cuf ~4 200 msec 100 msec S B P seen at proximal edge of cuff =120 mmHa ^ PWTT = 376 msec w I I cuff _—-',25 m s e c * S B P S e e n 3 t P r 0 x i m a l edge of cuff = 90 mmHg PWTT = 310 msec S B P seen at proximal edge of cuff = 90 mmHg 200 msec •X • 110 msec Elevation of limb results in a drop in blood pressure as measured on the elevated limb but the change in PWTT is not proportional to that seen due to an equivalent change in overall systemic blood pressure since PWV only changes over the elevated section of the arterial path Fig. 3.9: The effect of limb elevation on blood pressure and PWTT. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 50 System to Allow Estimation of LOP for Tourniquet-Related Usage CHAPTER 4 THE DESIGN AND IMPLEMENTATION OF A PROTOTYPE PWTT MEASUREMENT SYSTEM TO ALLOW ESTIMATION OF LOP FOR TOURNIQUET-RELATED USAGE 4.1 System Overview In order to allow clinical analysis of the relationship between changes in PWTT and changes in BP it was necessary to build a PWTT measurement system. For reasons discussed previously it was decided to build a system which measures the elapsed time between the occurrence of the R-wave in the ECG waveform and the arrival of the resulting pulse at a peripheral point. It is expected that the ECG signal can be obtained from the analog output of the ECG monitor typically utilized during surgery while the pulse signal is obtained from a dedicated sensor. The measurement system was developed using an Intel 20MHz 80C196 16-bit microcontroller to perform the necessary signal processing, control and calculation functions. This microcontroller is a register-to-register machine and most operations can be quickly performed from or to any of the registers. The register operations control the many peripherals which are available on the chip including a serial port, A/D converter, PWM output, up to 48 I/O lines and a high speed I/O subsystem which has two 16-bit timer/counters, an 8-level input capture FIFO and an 8-entry programmable output generator. The 80C196 is equipped with 230 bytes of RAM. A block diagram of the measurement system is given in Fig. 4.1. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 51 SAMPLING CONTROL ARTERIAL PULSE DETECTOR GAIN CONTROL ECG MONITOR FILTERING R-WAVE r DETECTION PWTT CALCULATIONS PWTT OUTPUT SYSTEM STATUS OUTPUT Fig. 4.1: Block diagram of the prototype PWTT measurement system. The tasks to be performed by the measurement system can be broken down into the following main items: • interface with the analog output of the an ECG monitor and pass the ECG signal through a detector which identifies the occurrence of the R-wave; • obtain an arterial pulse signal indicative of the arterial pulse sensed by a probe located at a peripheral point on the body; • pre-condition the arterial pulse signal to remove noise and corruption due to motion artifact; • perform analysis of the arterial pressure signal to identify noise or motion artifact that is not removed by the filter stages which might inhibit the accurate detection of the pulse arrival; • perform analysis of the arterial pressure signal to detect the arrival of the leading edge of the arterial pulse; Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 52 System to Allow Estimation of LOP for Tourniquet-Related Usage • calculate the elapsed time between the detection of the R-wave and the leading edge of the resulting arterial pressure pulse; • calculate a running average of the transit time values to smooth out beat to beat variations in PWV and output the resulting value to a user interface; and • provide feedback to the user regarding the quality and magnitude of the ECG and arterial pressure waveform signals and the results of the noise detection routines. Detailed schematics of the system and software listings are included in Appendices A & B. 4.2 Detection of the R-wave from the ECG Signal A great deal of study has been undertaken with regard to the analysis of ECG waveforms including much work involving automated timing of interbeat intervals to detect abnormal rhythms which indicate various heart conditions. As a result, there are many proven R-wave detection hardware designs published [47-51]. The output of such a circuit can be utilized in conjunction with the high speed input ports of the 80C196 microcontroller to provide a trigger for the start point of the PWTT timing cycle. It was therefore determined that the use of a hardware based detector rather than a software algorithm would be most suitable for implementation in the prototype system. The ECG R-wave is relatively uniform among individuals in the sense that it can be easily identified as a large spike in the ECG waveform which is significantly greater in magnitude than any other periodic feature in the cardiac cycle. Most hardware based R-wave detector designs appearing in the literature utilize a filter that selects the frequencies present in the QRS complex and then trigger using some kind of amplitude thresholding circuit [47]. The desirable passband for the filter to maximize QRS energy has been found to be centred at Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 53 System to Allow Estimation of LOP for Tourniquet-Related Usage approximately 15-17 Hz [47,48,49,50,52]. More complex detection schemes such as adaptive filtering [50] or 1 st and 2nd derivative based thresholding [49] have proven effective in ambulatory monitoring where noise due to poor electrode application, motion artifact and muscle noise may make detection more difficult. However, given that relatively stable, high quality ECG signals are normally obtained from patients during surgery and since subjects undergoing surgical procedures under general and regional anesthetics can be expected to remain relatively still limiting motion artifact in the ECG signals, it was concluded that a scheme based on amplitude thresholds would provide simple and clinically adequate results for this application. The R-wave detection hardware was developed using signals from an Electronics for Medicine model PM-2A ECG monitor (Electronics for Medicine, Pleasantville, N.Y.), typical of the ECG machines utilized in the surgical rooms of Vancouver Hospital and Health Sciences Centre. An analog sample of the ECG waveform is available a 2-pin output jack at the back of the device. This is typical of most stand-alone ECG monitors. The ECG monitor produces a signal of approximately 1-3V peak-to-peak amplitude depending on the magnitude of potentials measured by the applied electrodes and the signal gain as set by the user controls. The R-wave detection circuit utilized for the prototype system shown in Fig. 4.2 is based on one proposed by N.V. Thakor [47,51], which was modified to remove the pacer rejection circuit at the input stage and to change the length of the output pulses to suit the PWTT measurement system. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 54 »5V 100K , "T" ECG SIGNAL . INPUT 1 ECS ANALOG , OUTPUT GROLND1 Fig. 4.2: R-wave detection circuitry. The R-wave detection circuit functions as follows. The ECG signal is passed through a bandpass filter of centre frequency 17 Hz and a Q of 3.3. A half-wave rectifier is utilized to provide an always positive output which is then compared to a threshold level set by an automatic thresholding circuit. This allows reliable R-wave triggering for a variety of QRS complex morphologies in the presence of motion artifact or other interference. The threshold varies automatically according to the sum of: (1) the filter output caused by most baseline interference and P-waves; and (2) 60% of the peak value of the previous R-wave. The sampled value decays with a time constant of 10 seconds. A 10 ms pulse is used to trigger the charging of the sample and hold circuit and a 200 msec re-triggerable monostable multivibrator causes a refractory period after the detection of each R-wave to prevent possible Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 55 System to Allow Estimation of LOP for Tourniquet-Related Usage triggering from the ringing output of the filter or triggering on other peaks in the cardiac waveform such as T-waves. The output of the circuit is a 200 msec, +5v pulse, the leading edge of which indicates the onset of the R-wave, defined as the point where the wave is 60% of the peak magnitude of the previous cycle. The circuit was tested in the laboratory on 4 subjects and found to work reliably, even in the presence of significant noise. Given the relatively constant amplitude of O.R. ECG signals and the fast rise time of the Q-R portion of the waveform (approximately 20-30 msec) the detector provides triggering at a point on the QRS wave that is sufficiently "constant" to allow accurate determination of PWTT changes. It was estimated through laboratory analysis of the ECG and detector signals from the 4 subjects using an oscilloscope and chart recorder that the error in the detection circuit is no greater than +/- 5 msec. The detector was found to operate effectively on signals over the 1-3V p-p amplitude as available from the ECG output and therefore no additional amplification stages were utilized to condition the ECG signal prior to input to the detection circuit. A protection circuit as recommended in the microcontroller manufacturer's reference literature consisting of two diodes, a 4700 pF capacitor and a 301 D. resistor was utilized to ensure that the voltage at the high speed input pin remains in the range as required by the 80C196. Should the input voltage inappropriately drop significantly below ground, the ground side diode will forward bias at about -.8v dc and if an over-voltage condition occurs, the reference side diode will bias such that the voltage at the pin is about 5.8 v dc. The resistor is utilized to limit input current during over-voltage conditions and the capacitor provides low pass filtering of the input to remove high frequency noise. This circuit is typical of that utilized on all inputs to the system 80C196 microcontroller. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 56 System to Allow Estimation of LOP for Tourniquet-Related Usage 4.3 The Arterial Pulse Sensor 4.3.1 Operational requirements The development of a sensor to detect the arrival of the arterial pressure pulse at a peripheral point on the body presents a significant engineering problem. For experimental purposes it was desired to built a sensor that is capable of detecting pulse signals from not only the fingers and toes but also the upper arm, thigh, ankle and forearm. A sensor has several key requirements for use in an adaptive surgical tourniquet system: • the sensor must be able to be applied quickly and easily in a secure manner by non-technical personnel and require little or no adjustment or repositioning to obtain a usable pulse signal; • the sensor must be relatively resistant to signal corruption due to hammering and chiseling of bone, the use of pneumatic oscillating saws or drills, or manipulation of the limb, all of which may occur during the course of an orthopedic procedure • the sensor must be capable of detecting pulse arrival in a wide variety of patients ranging from those with light skin tone and excellent blood perfusion to those with dark skin tone and very poor perfusion; • the sensor must be impervious to liquids, unaffected by physical shocks encountered in use, able to stand up to the physical stresses of repeated usage in the operating room and have stable sensitivity over time; and • the sensor should have a cost to manufacture similar to the cost of sensors used for pulse oximetry which are commonly used. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 57 System to Allow Estimation of LOP for Tourniquet-Related Usage 4.3.2 Available technologies for pulse sensing While it was eventually decided that a photoplethysmographic sensor was most suitable for use in the prototype measurement system a thorough survey of potential sensors was made. These are discussed in the following paragraphs. The appropriate use of strain gauges may allow the detection of the flow of blood into a limb by measuring movement of the underlying tissue. One possible configuration is proposed by Hokanson et al [53] in which an electrically calibrated mercury strain gauge is attached to rubber bands which encircle the limb. Variations in limb circumference related to the volume of blood in the underlying tissue can then be measured in terms of increased tension on the strain gauge. However the clinical use of strain gauges is impractical due to their extreme sensitivity to motion artifact and because strain gauges sensitive enough to detect the small movements signifying blood flow are highly susceptible to damage from rough handling or being dropped. In addition, strain gauges utilized to detect pulsatile blood flow require careful application to ensure that the tension in the bands encircling the limb is correct; too tight and the sensor could be damaged, too loose and no signal will be obtained. It is also necessary that the sensor be applied with uniform tension around the circumference of the limb. Electric impedance plethysmography based on impedance changes of the tissue between two electrodes in response to the pulsations of blood is not suitable since the accuracy of the technique has been found to be poor [37]. Ultrasound or tonometric methods which require that the sensor be placed directly over an artery were also considered unsuitable because they are highly dependent on correct positioning and as a result such sensors are difficult to apply consistently. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 58 System to Allow Estimation of LOP for Tourniquet-Related Usage Laboratory experiments involving the use of pressure sensors to detect the pressure change in the tourniquet bladder as the arterial pulse wave "hits" the occluded arteries underlying the tourniquet cuff met with varying success. The use of pressure sensors placed directly on the skin under the cuff or attached to a small liquid or air filled bladder located under the cuff, while relatively resistant to motion artifact, generally produced only weak pulse signals. The use of a pressure sensor to monitor the pressure within the surgical tourniquet cuff bladder itself produced much better results as far as signal strength but was found to be highly sensitive to artifact since any slight motion of the limb produced large pressure fluctuations in the tourniquet cuff. In both cases the pulse shape tended to be poorly defined compared to that obtained with other techniques, thus increasing the difficulty of determining the exact point of pulse arrival. 4.3.3 The photoplethysmographic sensor Photoplethysmography (plethysmography meaning the measure of changes in volume - in this case the volume of blood) utilizes the light absorption properties of blood to allow measurement of the relative volume of blood in vascular tissues. The technique involves the placement of a light source and a photodetector over or on opposite sides of a vascular bed and measuring the changes in the percentage of incident light from the source reaching the photodetector. The light level seen at the photodetector consists of two components. The dc "baseline" depends mainly on the total volume of venous, non-pulsatile blood in the region, the opacity of the skin, reflection of the incident light by bones, and tissue scattering [54]. The ac "pulse signal" which rides on top of the baseline, corresponds to the pulsatile flow of blood through the arterial side of the capillary bed. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 59 Fig. 4.3: The absorption of light in the vascular bed [51]. The incident light on a can be described in terms of the following equation: 'incident = 'transmitted + Ubsorbed + 'backscattered (eq- 4.1) Looking at the portion of the incident light which is backscattered, I backscattered, the magnitude of this quantity then varies as a function of the properties of the underlying tissues, •backscattered — f(A,issue) + f(Avenous blood) + f(Aarterial blood) (eq. 4.2) where A « „ u e , Avenous wood, and Aa r t eriai biood are the absorption due to tissue, venous blood and arterial blood respectively. While the absorption of the incident light due to the underlying Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 60 System to Allow Estimation of LOP for Tourniquet-Related Usage tissue and the volume of venous blood will remain essentially constant over the cardiac cycle, the absorption due to arterial blood will vary as a function of the time varying volume of blood in the arterioles. Aanerial b | 0 0 d may then be defined as, Aarterial blood = Anormal blood volume A p u | s a t j | e blood volume (eq. 4.3) where A n 0 rmai biood volume is the absorption due to the constant volume of the arterial blood in the capillary bed and A p U | Satiie wood volume is the pulse added volume of blood which corresponds to the arrival of each arterial pulse wave. Depending on the physiological state of the microvascular bed, this alternating light intensity due to the pulsatile flow of blood amounts to approximately 0.05% - 1% of the total light intensity either transmitted through or backscattered from the skin [54]. Changes in oxygen content of the blood also affects the percentage of light absorbed; this is the principle on which pulse oximeters which measure blood gas content are based. However, the resulting changes are small and essentially dc when compared to the signals due to pulsatile blood flow. Photoplethysmographs or photoelectric sensors can be either reflectance or transmittance type design. Transmission sensors are generally used on the fingers or toes, the nasal septum or the earlobe. While it is possible to use transmission sensors on a fleshy fold of skin this method of application is not particularly reliable since the detector is liable to fall or be accidentally pulled of over the course of an operation. It is a simple matter however to detect the case that the sensor becomes dislodged. Chapter* The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 61 REFLECTANCE TYPE INCIDENT LIGHT IS REFLECTED FROM UNDERLYING TISSUE AND BONE TRANSMISSION TYPE PHOTODIODE INCIDENT LIGHT FROM LED PASSES THROUGH FINGER Fig. 4.4 - Reflectance and transmittance type photoelectric sensors Practically, the major limitation in reflectance sensor photoplethysmography is the low level pulsatile signals typically recorded from low density vascular areas of the skin. The feasibility of the use of a reflectance type pulse probe is therefore dependent on the ability to design a sensor that can detect sufficiently strong signals from the intended locations on the body. The following sections detail the main considerations in the design of a photoplethysmographic sensor. LED Current The light emitted from a LED is a function of the driving current and therefore the driving cunent determines the effective penetration depth of the incident light and thereby the amount of the vascular bed which will be illuminated. In order to maximize the pulsatile signal seen at the photodiode, the LED driving current should be as high as practical without saturating the photodiode. LED's sufficiently small for use in a sensor are only capable of passing a few mA continuous current and therefore in order to utilize sufficiently high driving current to Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 62 System to Allow Estimation of LOP for Tourniquet-Related Usage adequately illuminate the vascular bed it becomes necessary to pulse the LED utilizing short duty cycle. Multiple LED's can be used to further increase the depth and range of vascular illumination. In using multiple LED's and large pulsed driving currents care must be taken to avoid heating of the surrounding tissues which might result in burns [55,56]. It is generally accepted that temperatures of no more than 40°C can be applied for durations of greater than 3 hours without danger of burn injury to the underlying skin and tissue [54]. The allowable heating of patient applied sensors is closely regulated by the applicable CSA safety standards. Photodiode Surface Area The amount of backscattered light detected by the sensor is proportional to the surface area of the photodiode. Where there are multiple diodes arranged at equal distances around the LED(s) the light detected by the probe is proportional to the number of photodetectors. LED Wavelength In the selection of LED wavelength it is necessary to choose a wavelength which has sufficient penetration into the tissue to reach the vascular bed, is sufficiently absorbed and scattered by blood to produce adequate pulsatile modulation, and is not so greatly absorbed by blood that the reflected signal is weak and hard to detect. Traditionally, infrared wavelengths (around 940 nm) have been utilized with excellent results however recent work by Cui and Ostrander [57] suggests that the use of light in the green portion of the spectrum (around 570 nm) might provide better signal modulation. The following plots summarize their findings. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 63 10 Vb • 0.0 S -k / I w . 1 , 1 . 1 , 1 , soo 700 800 Wavalangth (nm) Pulaatlla Blood Voluma • 0.0S Vb Wavalangth (nm) The left figure gives scattering and absorption coefficients (t. w) of tissue used in the model. The value of k is independent ci( blood 0 , saturation and is therefore a single curve for all saturation conditions. On the right is the calculated reflectance modulation for pulsatile blood. The model results below 500 nm are not valid, since the photon diffusion theory assumptions arc not met in this range. Curve Arterial blood 0 2 saturation Skin (issue blood 0 , saturation Dashcd-Solid Dashed dolted 97% 70% 0% 80% 60% 0% Fig. 4.5: Scattering and absorption coefficients of light in human tissue [57]. These findings suggest that maximum pulsatile modulation, defined as the ratio of the pulsatile variation to the mean reflected light intensity, is some 7X greater at the green portion of the spectrum than at the infrared. To investigate these findings further, a pulse sensor utilizing both a green wavelength LED and infrared LED which could be driven independently was constructed and tested as part of this project. In practice the use of the green wavelength LED was found to be greatly inferior to an infrared wavelength LED. While as Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 64 System to Allow Estimation of LOP for Tourniquet-Related Usage expected the green light showed greater modulation as a percentage of the dc signal, the overall signal strength was far smaller than that obtained from an infrared LED driven with the equivalent current and was poorly suited for obtaining an arterial pulse signal for pulse wave arrival detection. Green light has a smaller penetration depth into the tissue than does infrared light and the pulsatile modulation from the green LED sensor therefore represents the pulsations of the blood in the dermal layer of the tissue. The modulation of an infrared LED sensor is dependent on blood volume both in the dermis and the underlying vascular tissue. The shallow depth of penetration with green light may prevent usable signals from being obtained from patients with poor perfusion. The difference in the depth of penetration of the incident light may explain the finding of the green LED being slightly more resistant to motion artifact than the infrared LED as the dermis may be less affected by limb movement than the deeper tissues. In addition, application pressure was found to dramatically affect performance in a green light sensor with the pulsatile signal diminishing drastically as the application pressure was increased. Cui and Ostrander found that skin colour does not appear to affect the shape of the modulation spectrum to any great extent since the pigmentation of the skin occurs only in the epidermis where no blood supply exists [57]. The ac modulation of the reflected light occurs only in the tissue beneath the epidermis which contains the blood. The epidermis proportionately absorbs both the mean intensity and the pulsatile component when light passes through it leaving the modulation unaffected but the entire signal attenuated. Provided that an adequate LED current is utilized, this can be compensated for by appropriate gain adjustment Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 65 System to Allow Estimation of LOP for Tourniquet-Related Usage Contact Pressure While it is difficult to precisely quantify in the general case, applying a sensor equipped with an infrared LED with a pressure of 20-50 mmHg was found in laboratory experiments to result in a stronger signal than simply contacting the probe with the skin. Greater contact pressures result in diminished signal strength due to the restriction of blood flow to the underlying vascular bed. Photodiode - LED Separation The intensity of the backscattered light decreases in direct proportion to the square of the distance from the LED's and therefore the distance between the photodetector and the LED's will significantly affect signal strength. Increasing the separation results in increased pulsatile modulation due the longer vascular path, but decreases the overall signal strength. It has been found by previous studies that a separation of 4-5mm provides the best sensitivity in terms of detecting adequately large pulsatile signals [58] however this is very much dependent on the radiating pattern of the LED's and the capillary perfusion of the underlying area. For poorly vascularized areas such as the thigh the use of a longer path length might allow detection of the pulse signal where a shorter path length might not. In laboratory experiments carried out as part of this project it was found through trial and error that a separation of 9mm provided the best performance for the configuration utilized for the experimental sensor. Photodiode Response It is necessary that the photodiode-LED pair be matched such that the photodiode sensitivity is high in the wavelength range of the emitted light. Ideally the photocell should have a Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 66 System to Allow Estimation of LOP for Tourniquet-Related Usage response in a narrow band around this frequency in order to help reject signals from ambient or background light, particularly 60 Hz fluorescent fixtures. Effect of Skin Temperature It has been found that local skin heating can be employed to produce vasodilation in the microvascular bed which can dramatically increase the pulsatile component of photoplethysmograms [54,60]. Mendelsen and Ochs [54] found that increasing local skin temperature from 34° to 45°C resulted in a fivefold increase in the pulse amplitude of infrared photoplethysmograms. They also found that this vasodilation caused the photoplethysmograms to become more stable, resulting in smaller beat to beat amplitude fluctuations. As noted previously care must be taken in prolonged heating of skin tissue to avoid burn damage. It is generally accepted that temperatures of no more than 40°C can be applied for durations of greater than 3 hours without danger of burn injury to the underlying skin and tissue [54]. The allowable heating of patient applied sensors is closely regulated by the applicable CSA safety standards such that no patient applied part can have a surface temperature of greater than 41 °C [59]. 4.3.4 The experimental sensor Based on consideration of the above factors a photoplethysmographic pulse sensor was designed and fabricated as shown in Fig. 4.6. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage SECTION DRAWING SECTION CENTRONIC BPW34 INFRARED FILTERED PHOTODIODE. TYPICAL OF TWO ACTIVE ELECTRONICS 940 nm INFRARED LED HARD RUBBER FACE PLATE WITH CUTOUTS TO RECESS LED AND • DIODES APPROX. 1mm. RECESSES FILLED WITH CLEAR EPOXY TO PROVIDE LIQUID PROOF SEAL FOR WIRING PULSE SENSOR - FRONT VIEW Features: single infrared LED, peak wavelength 940 nm LED voltage 1.6v, max. power dissipation 100mW, max. continuous current 50mA LED current utilized = 400 mA @ 2.6% duty cycle ( a function of the current driver circuit) two Centronic BPW 34 photodiodes, total active surface area 14.6 mm2, photodiode spectral response peak @ 950 nm centre-centre LED photodiode separation 9 mm LEDs and diodes recessed 1mm, window dimensions approx. 5mm X 5mm insulating rubber faceplate maximum thickness 4 mm, outer dimensions 21 mm(w) X 29 mm(l) opaque backing to block ambient light strain relief for lead wires all terminal connections sealed with waterproof, non-conducting epoxy secured to finger or toe using Velcro strap which encircles digit and blocks ambient light Fig. 4.6: The experimental photoplethysmographic pulse sensor. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement QQ System to Allow Estimation of LOP for Tourniquet-Related Usage The sensor was constructed using a single infrared LED pulsed at 3 msec intervals utilizing a driving current of 400 mA. Each LED pulse is 80 usee in duration yielding a duty cycle of 2.6% and a time-averaged power consumption on the order of 17 mW; 17% of the rated power dissipation for the LED or 21% of the maximum power rating based on the rated maximum continuous driving current. While no formal testing was carried out to determine the heating of the arterial pulse sensor, the sensor was applied for periods of up to several hours in extensive laboratory testing during which at no time was noticeable warming of the sensor LED noted. In addition, given that the LED is operating at 17% of the rated power dissipation, no significant heating should occur. The LED current source is fused in a manner designed to interrupt the LED current in the event that a current in excess of the design current is applied. The magnitude of the photodiode response was found to increase linearly as the driving current of the LED was increased to this level and it was therefore determined that use of currents up to 400 mA caused no reduction in the intensity of illumination as a function of driving current. For LED currents higher than 400 mA the photodiode went into saturation in some applications of the sensor. The current for driving the LED was provided from a bi-directional current source as shown in Fig. 4.7. This bi-directional driver circuit was originally designed to allow the use of two LED's in order to test the response of the photodiode to different LED's. For the final implementation the driver is used to provide only positive current to drive the single LED as described. The magnitude of the driving current is adjusted by the setting of the 10KT2 potentiometer shown in Fig. 4.7. Switching of the LED is carried out in response to +5v trigger signals from the high speed output port of the microcontroller described in detail later in this chapter. As shown in Fig. 4.7 two 1000uF capacitors are connected in series with the emitters of the matched transistors in order to provide a source for the 400mA current pulses and to Chapter 4: The Design and Implementation of a Prototype PWTT Measurement gg System to Allow Estimation of LOP for Tourniquet-Related Usage limit the switching noise resulting at the main power buses. In spite of this precaution some switching noise does occur. Fig. 4.7: Circuit for the LED current source. The two photodiodes depicted in the experimental sensor shown in Fig. 4.6 are utilized in order to provide a large surface area for detection of the reflected light and increase the effective vascular bed over which the blood volume measurement is taken. Centronic BPW34 photodiodes (Centronic; Newbury Park, CA) were chosen due to their compact size relative to their effective surface area and due to their excellent sensitivity, low capacitance and 350 nsec response time. The photodiodes are each centred 9 mm from the LED, a distance Chapter* The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 70 which was decided upon by trial and error after experimenting with separations ranging from 3-15mm. The photodiodes are connected in parallel and the current induced in the photodiodes by incident light is converted to a representative voltage utilizing a simple op-amp biasing circuit which feeds the photocurrent to an operational amplifier virtual ground as shown in Fig. 4.8. The generated photocurrent flows through the biasing resistor and the resultant voltage at the output of the op-amp is linearly dependent on the incident illumination level. The biasing circuit shown in Fig. 4.8 generates little noise due to the almost complete elimination of leakage current [61]. Reverse biasing the photodiode with a positive or c negative voltage with respect to ground would allow faster response than the zero bias circuit; however there is an increase in leakage current due to the biasing voltage which can result in significantly higher noise levels. 11pF SENSOR PHOTODIODE CMOS ANALOG a 2 K CMOS ANALOG SWITCH AJ^ | SWITCH |-VVV-| | - Q AM8IENT LIGHT REJECTION CONTROL HSO.2 fj SAMPLE AND HOLD CONTROL HSO.1 TO AMPLIFICATION AND ANALOG RLTER STAGES PHOTODIODE BIASING STAGE AMBIENT LIGHT REJECTION STAGE SAMPLE AND HOLD STAGE Fig. 4.8: Circuit for biasing and sampling of the photodiode signals. The resulting voltage signal is a summation of the response from the ambient light incident on the photodiodes and the response due to the reflected light from the pulsed LED. The sensor Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 71 System to Allow Estimation of LOP for Tourniquet-Related Usage housing is designed to minimize the contribution of ambient light and an ambient light rejection circuit is utilized which references the response of the signal from the LED pulse against a capacitor allowed to charge to the voltage output in response to ambient light levels. A small amount of ringing is also noted with the photodiode response to the LED pulse however the addition of a small (11 pf) capacitor across the biasing resistor shown in Fig. 4.8 was found to effectively filter out much of this ringing by attenuating the high frequency components associated with this response. The attenuation of these high frequency harmonics results in a slight rounding of the rising and falling edges of the pulse waveforms. The continuous photoplethysmographic waveform is reconstructed from discrete samples taken at times corresponding to the LED pulses. These samples are timed to occur during the middle portion of the photodiode voltage pulse in order to minimize the effects of ringing and switching noise. The timing of the LED and sampling events is controlled through the high speed output port of the microcontroller and shall be described in detail later in this chapter. No attempt was made at integrating a heater into the sensor as part of this project. Since heating of the sensor to temperatures greater than 40°C carries risk of patient injury [54] and heating to temperatures greater than 41 °C is prohibited under CSA standards [59], a heating system for use in the sensor must allow accurate temperature control and is therefore a significant undertaking to develop. Temperature measuring chips are prone to hysteresis and small heating elements are somewhat difficult to accurately control to achieve a steady temperature level. It was therefore decided that for this project the potential gains in signal strength could not justify the development of a heating element. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 72 System to Allow Estimation of LOP for Tourniquet-Related Usage The sensor shown in Fig. 4.6 was constructed to be as thin as possible to facilitate application under the proximal edge of a tourniquet and the sensor was equipped with a 3m lead to allow application with minimum intrusion into the operating area during clinical trials. The wires carrying the photodiode current are isolated from the LED current leads by a braided copper shield. Strain relief is provided at the point of connection between the lead and the sensor housing. The faceplate of the sensor was fabricated from insulating rubber into which the LED and photodiodes are recessed and all wire terminations and exposed contacts on the sensor were sealed in non-conductive epoxy to electrically isolate the sensor from the patient. A velcro strap was designed which allows the sensor to be easily applied to the finger or toe and to shield the sensor from ambient light. 4.3.5 Modification of the sensor for application under the tourniquet cuff For application of the pulse sensor at the location of the tourniquet cuff it may be of significant advantage to locate the sensor just under the proximal edge of the tourniquet cuff as shown in Fig. 4.9 rather than simply taping or otherwise fastening the sensor to the surface of the skin. It was found in laboratory experiments performed on various subjects that locating the centre of the sensor about 1 inch under the proximal edge of the tourniquet cuff produced the optimum signal although this distance was found to vary greatly depending on the cuff application and pressure. The signal obtained from a sensor applied under the cuff may benefit from the pressure pulse in the cuff bladder caused by the pulse wave hitting the occluded point in the arterial path. It was found in several cases during laboratory testing that an identifiable pulse was seen using the probe under the distal half of the cuff while the tourniquet was inflated to supra-occlusive pressures. Since in this case there is no pulsatile flow in the capillary bed underlying the sensor, it is suggested that the "pulsing" of the tourniquet causes a small shift in the probe relative to the underlying tissue which results in a Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 7 3 System to Allow Estimation of LOP for Tourniquet-Related Usage modulation of the signal. As this pulse is in phase with the modulation due to blood flow, the addition of these two effects may provide improved signal amplitude as compared to simply affixing the sensor to the limb using tape or similar means. When a tourniquet cuff is removed after being applied to a limb for more than a few minutes there are red marks and depressions on the skin at any point where there was a discontinuity in the pressure profile such as at seams, where the cuff overlapped, or where the cuff buckled as it was inflated. Even the stitches in the trim will leave readily identifiable marks. It is common practice to use a stockinet under the tourniquet cuff in order to provide a cushion and limit marking. Applying the photoelectric pulse sensor directly under the cuff without some sort of protective housing will result in a large depression being left in the limb which is not acceptable in clinical use. After experimentation with a number of designs it was found that by utilizing a clear, viscous silicone gel filled housing between the sensor and the limb, the sensor could be applied under a cuff with no appreciable marking, even at pressures as high as 400 mmHg. The housing shown in Fig. 4.9 was developed to allow application of the sensor under a tourniquet cuff. Applied in this housing, the sensor floats on top of the gel mass which conforms to the shape of the tourniquet and the underlying tissue such that no pressure discontinuities occur on the limb surface that can cause blood pooling. The only tissue marking results from the discontinuity caused by the thickness of the layers of polyethylene which causes a small ring shaped mark to be left around the edge of the gel sack. This marking is no greater than that normally caused by the tourniquet seams or the overlap of stockinet wrapping and is therefore considered acceptable. For clinical testing, a velcro strap Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 74 System to Mow Estimation of LOP for Tourniquet-Related Usage was affixed to the gel housing such that it could be secured a fixed distance under the tourniquet. PROXIMAL * • DISTAL LIMB 1-3 cm SENSOR LEAD LOCATING THE SENSOR UNDER THE TOURNIQUET CUFF 60 mm sensor lead membrane heat sealed to form a liquid-tight sack GEL FILLED SENSOR HOUSING Fig. 4.9: Gel housing for application of the arterial pulse probe under the proximal edge of a tourniquet cuff. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 7 5 System to Allow Estimation of LOP for Tourniquet-Related Usage 4.4 Pre-conditioning of the Pulse Signal 4.4.1 Reconstruction of the pulse signal Prior to reconstruction of the arterial pulse signal from the pulsed LED samples, signals induced by ambient illumination are rejected by a unity gain differential amplifier circuit as shown in Fig. 4.8. The photodiode signal is input to a sampling circuit controlled by a pin of the microcontroller high speed output port which samples and holds the signal immediately prior to the time at which the LED is switched "ON". The output of this sample and hold circuit, equivalent to the photodiode response to the ambient light level, is fed to the inverting input of the differential amplifier. The photodiode signal is fed to the non-inverting input of the amplifier such that the amplifier output over the 80 usee duration of the LED pulse is a value which corresponds only to the portion of the photodiode signal resulting from the LED illumination. In order to reconstruct the photoplethysmographic waveform from the response of the photodiode to the pulsed LED, the signal is passed through a sample and hold circuit triggered by a pin of the microcontroller high speed output port. The signal is sampled at 3 msec intervals at a point during the middle of each LED pulse to avoid corruption due to switching noise from the LED current driver and ringing of the photodiode output. Details of the timing of the high speed output port signals are provided in section 4.6.2. The reconstructed signal is a combination of the small signal due to changes in the blood volume of the vascular bed superimposed on the large dc output resulting from reflection of the incident light by bone, venous blood and soft tissue. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement JQ System to Allow Estimation of LOP for Tourniquet-Related Usage 4.4.2 Filtering Filtering is required to remove the dc component of the photoplethysmographic signal to isolate the pulsatile component, and to reduce the effects of noise due to motion artifact, circuit noise other sources present in the noisy environment of an orthopedic surgical procedure. Circuit diagrams for the filtering stages are provided in Appendix A. For a pulse sensor applied at the finger of a non-operative limb where little motion is likely to occur, the noise component of the signal is generally quite small. For a pulse sensor applied at the tourniquet cuff location or elsewhere on the operative limb the noise component is significantly higher, particularly during periods of limb manipulation or the use of hammers, chisels or pneumatic tools. While isolation of the pulsatile signal from the dc component can be accomplished through the use of a simple high pass filter, filtering to reduce signal distortion due to noise is more difficult. Given the wide range of arterial pulse shapes that occur depending on factors such as sensor location, limb position, skin temperature, heart rate, arterial tone and blood pressure which vary not only from subject to subject but within a given individual over the course of an operation it is impossible to characterize the shape of a "typical" noiseless arterial pulse signal. This is typical of many biological signals and as a result the use of use a matched filter is not practical as a means to isolate the desired pulsatile signal from background noise. Various schemes exist for the use of adaptive filters which attempt to isolate the frequencies characteristic to the arterial pulse of each individual case however these tend to be complex and place a high computational demand on the microcontroller. The algorithm developed for filtering the arterial pulse sensor signal relies on a combination of analog filtering and noise detection software and was developed in consideration of the following: Chapter 4: The Design and Implementation of a Prototype PWTT Measurement JJ System to Allow Estimation of LOP for Tourniquet-Related Usage (1) the information contained in the pulsatile signal is confined primarily to a relatively small frequency band ranging between approximately 0.5-20Hz; (2) broad spectrum "background" noise due will tend to be small in amplitude compared to the pulsatile signal; (3) noise due to the use of pneumatic tools will tend to be higher in frequency than the pulsatile information [21]; and (4) noise due to hammering, chiseling, rapid manipulation of the limb or similar corrupting factors will be intermittent and large in amplitude compared to the pulse waveform. Since a large portion of the noise which can corrupt the pulse sensor signal occurs outside of the frequency spectrum occupied by the arterial pulse waveform information, a simple cascaded low pass, high pass filter combination is used to reject frequencies outside the range of the pulse waveform. Noise can then still result in the signal band. However, given that such noise is generally large in amplitude and periodic, resulting from manipulation of the limb or hammering and chiseling of bone and since such intermittent large amplitude noise can readily be identified through software analysis, the measurement system then relies on the use of software error detection routines to detect the presence of such noise and reject the corrupted periods for PWTT calculations. Since the use of measured PWTT to estimate BP only requires a single noise-free cardiac cycle to produce a valid estimate it is not necessary to obtain a continuous noise free signal from the pulse sensor. The requirements for filtering of the pulsatile signal are different from most signal processing problems found in communications which require that all information present in the signal be accurately identified. Periods of excessive noise can be discarded and PWTT values calculated from clean periods without adversely effecting the acquisition of adequate PWTT information to Chapter 4: The Design and Implementation of a Prototype PWTT Measurement JQ System to Allow Estimation of LOP for Tourniquet-Related Usage allow adaptive operation of a surgical tourniquet system. The software algorithms for noise detection are discussed further in section 4.5.4. While there is some literature suggesting various filter parameters for arterial pulse signal processing the choice of cutoff frequencies for the filters was largely a result of trial and error using a variety of test subjects and pulse probe locations. Laboratory experiments showed that the bulk the arterial pulse information is generally contained within the frequency spectrum of about 0.1-20 Hz; however, this is highly dependent on pulse morphology. A three-pole Butterworth high-pass filter is utilized with a cutoff frequency of 0.25 Hz. This cutoff point was chosen as a compromise between increasing baseline wander by using a lower cutoff and excessive ringing caused by higher cutoff points. A four-pole Butterworth low-pass filter is utilized with a cutoff frequency of 20 Hz. The choice of the cutoff frequency for the low-pass filter stage was found not to be as critical as the high-pass filter stage, with low-pass corner frequencies between 15-30Hz producing similar results. The frequency response of the initial filter stages was confirmed experimentally using a sinusoidal signal generator and an oscilloscope to record the amplitude response over the relevant frequency range. The result is shown in Fig. 4.10. Samples of typical filtered and unfiltered signals are given in Fig. 4.11. Complete circuit diagrams for the filtering stages are provided in Appendix A. Chapter* The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage m -o L O W FREQUENCY RESPONSE Hz -a 3 m • a HIGH FREQUENCY RESPONSE 40 60 Hz Fig. 4.10: Frequency characteristics of input filter stages. B unfiltered filtered unfiltered filtered Fig. 4.11: Typical pre-filtered and post-filtered arterial pulse signals. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement QQ System to Allow Estimation of LOP for Tourniquet-Related Usage The photoplethysmograph signal, filtered and isolated from the dc component is then passed through a two stage adjustable gain voltage amplifier. The first stage provides a gain which can be adjusted in 16 steps between 1.5 and 42 by four CMOS switches. These can be controlled using the output lines of the microcontroller. The second stage has a fixed gain of 42 yielding an overall gain variable from 63 to 1764. An additional one pole low pass filter (15 Hz) and one pole DC blocking filter (0.25 Hz) are included between the two gain stages to remove stray DC offset and circuit noise which arises from the gain amplitude switching circuits. These will have only a small effect on the signal as compared to the multi-pole filters at the input stages. The resulting signal is biased at +3v to provide a 0-5v input to one of the eight channels of the 10 bit successive approximation a/d converter located on the 80C196 microcontroller. A overvoltage protection circuit as described previously is included at the input. Complete circuit diagrams for the input stages are included in Appendix A. 4.5 Software 4.5.1 System overview The software developed for the prototype PWTT measurement system was written in assembly code using a PC text editor. This assembly code was compiled using an Intel ASM-96 Batch compiler and downloaded to an Intel 196KD-20 Microcontroller Target Board via the PC serial port. The 196KD-20 Microcontroller Target Board emulates the operation of an Intel 80C196 microcontroller and provides a direct interface with the serial port of a PC allowing the use of associated programs running under Microsoft Windows™ to debug controller software. The final version of the developed software was burned onto an 8K EPROM which was Chapter 4: The Design and Implementation of a Prototype PWTT Measurement Q -| System to Allow Estimation of LOP for Tourniquet-Related Usage installed on the target board which was then configured to operate in "stand-alone" mode. When the reset button on the target board is pressed, the microcontroller begins execution of software commands located at a specified start address of the EPROM memory. A block diagram of the software is shown in Fig. 4.12. A complete listing of the system software is provided in Appendix B. 4.5.2 LED control and system timing All timing functions for PWTT measurement are clocked in terms of cycles of the 333 Hz LED timing routine which runs continuously in the background to pulse the LED and perform switching functions for the arterial pulse signal acquisition hardware described previously. PWTT values are determined in terms of the number of LED pulses (3 msec periods) between the positive transition from the R-wave detector circuit and the LED pulse sample at which the leading edge of the arterial pulse is detected. The LED timing routine serves to control turning the LED on and off, triggering of the sample and hold switch for the pulse signal, control of the ambient light rejection sample and hold switch, and instigation of the a/d conversion of the pulse sample. Sampling and a/d conversion times have been offset from the LED switching points to avoid transient noise which occurs due to switching as discussed previously. The 10-bit a/d conversion requires approximately 22 usee to complete and is undertaken on the sampled waveform after the completion of all other switching functions in each 3 msec cycle. Fig. 4.13 details the timing of these switching functions. Chapter* The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet Related Usage 82 S T A R T NTIAUZE VA BEGN ARTE SAW BABLES AND •HAL PULSE UNO STORE LATEST PULSE SAMPLE N STACK COmJTE SLOPE AM] ACCELERATION FROM LAST 3 PUJ3E SAMPLES GO TO PULSE ARRIVAL DETECTION ROUTINES LOOK FOR PULSE ARFSVAL POINT WAS AN ECG R-WAVE DETECTED SINCE LAST SAMPLE ? - Y E S - d CALCULATE PLLSE CHARACTERISTICS FROM CURRENT CYCLE CHARACTERISTICS MEET THRESHOLD P1XSE LED IN ORDER TO GO TO LED PLLSE ROUTINE OBTAIN A NEW FHOTOR-ETHl'SMOGflAPHC SAWLE R E T U R N Fig. 4.12: Block diagram of software routines. Chapter* The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 83 AMBIENT LIGHT S/H LED CONTROL SIGNAL S/H A/D CONVERSION TIME (usec)-0 8 LED ON 40 I S/H ON AMBIENT LIGHT S/H OFF 72 S/H OFF 88 LED OFF 120 I START A/D CONVERSION 138 TIME IN usee AMBIENT LIGHT S/H ON Fig. 4.13: Pulse timing functions. These switching events are timed accurately using the CAM (content addressable memory) registers. There are 8 CAM registers and one register is compared to the system internal timer values every state time to determine if the associated action is to be carried out. This defines the time resolution of the high speed output (HSO) to be 8 state times or 0.8 usee given the 20 MHz oscillator. Each CAM register is 24 bits wide -16 bits specifying the time at which the action is to be carried out and 7 identifying the nature of the action to be carried out and the reference timer to be utilized (two timers are available within the microcontroller). 4.5.3 Pulse arrival detection routine In detecting the arrival of the arterial pressure pulse, beat-to-beat shape and amplitude changes make using slope or magnitude thresholding unsuitable since the jitter caused by Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 84 System to Allow Estimation of LOP for Tourniquet-Related Usage pulse morphology changes could easily dwarf the PWTT changes expected to indicate shifts in blood pressure. Given the wide range of arterial pulse shapes that occur depending on factors such as sensor location, limb position, skin temperature, heart rate, arterial tone and blood pressure which vary not only from subject to subject but within a given individual over the course of an operation it would be extremely complex to characterize the shape of a typical arterial pulse and therefore the use of shape matching or pattern templates to identify arrival is also unsuitable. Benthin et al [62] found the estimation of PWV from cross correlation of successive regional pulse wave recordings to be subject to error due to phase shifts caused by reflections in the arterial tree. They found the magnitude of these reflections to be highly dependent on the position of the pulse probe and to vary significantly between subjects and suggest that cross correlation based PWTT measurement not be undertaken "too" close to any expected source of reflection - such as the tip of a finger or a tourniquet. Although the shape of the pulse is highly variable, one feature of the pulse shape was found to be suitable for measurement of arrival time: the first sign of upturn as the leading edge of the pulse wave reaches the vascular tissue beneath the sensor. While the pulse waveform may be subject to varying distortion due to reflected waves, this should not affect the identification of the pulse arrival point using a leading edge detection scheme. One possible technique for detection of the leading edge is to simply look for the last zero or negative slope prior to the rise in the waveform leading to the pulse peak. The approach depicted in Fig. 4.14 was found to provide excellent results in the majority of cases involving tests on 5 normal subjects, although it was found that some intelligent filtering must be utilized to remove jitter from the calculated slope waveforms such that small spikes in the waveform during the pulse rise do not repeatedly trigger the detector. Chapter* The Design and Implementation of a Prototype PWTT Measurement System to Mow Estimation of LOP for Tourniquet-Related Usage 85 PEAK MAGNmjDE Fig. 4.14: Detection of the leading edge of the pulse through slope upturn. Often, an arterial pulse waveform was obtained which has the characteristic of the signal slowly increasing prior to the arrival of the pulse wave. Using a slope based detection technique as shown in Fig. 4.14 on such waveforms the arrival of the leading edge will be determined to be at a point much earlier than is the actual case. This is shown in Fig. 4.15. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 86 peak magnitude Fig. 4.15: Potential error in the use of slope upturn detection. A better technique to detect the arrival of the arterial pulse and that utilized in implementation of the system software is based on identifying the peak acceleration point in the pulse period. As seen in Fig. 4.16, the arrival of the leading edge coincides with a peak in the second derivative of the waveform. This peak is not significantly affected by the slope of the waveform prior to arrival of the leading edge but depends essentially on the occurrence of a rapid upward deflection in the waveform as the pulse energy reaches the sensor. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 87 Fig. 4.16: Detection of arrival of leading edge by peak acceleration. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement QQ System to Allow Estimation of LOP for Tourniquet-Related Usage Using the acceleration peak to identify the onset of the pulse wave, pulse arrival can be detected by monitoring slope change to identify upturned portions of the waveform over which the point of peak acceleration is then found. Given that in a signal uncorrupted by large magnitude noise the peak of the arterial pulse wave will represent the maximum amplitude of the waveform over each cardiac cycle, pulse arrival can subsequently be identified as the point of peak acceleration on the upturned section of waveform leading to the point of maximum amplitude. Utilizing this amplitude criteria removes the danger of small noise spikes, which may dominate the second derivative waveform, from being mis-identified as the arrival point. Amplitude threshold limits are also utilized in the detection algorithm which prevent the selection of acceleration peaks from short upturned waveform segments near the pulse peak. These can occur as a result of small amplitude jitter which can cause positive negative slope transitions as the waveform flattens near the pulse crest. Fig. 4.17 shows a block diagram of the leading edge detection algorithm as implemented in the system software. In order to provide additional resistance to false triggering from high frequency amplitude jitter or noise spikes, the slope and acceleration values utilized for comparison are calculated using the current sample value and that 3 and 2 samples back respectively. This has effect of attenuating derivatives calculated from magnitude changes that are short in duration ( < 9 msec ) as compared to sustained changes as would normally be seen in the arrival of the pulse. This is shown in equations 4.4-4.6: CURRENT PULSE MAGNITUDE = M(T„), (eq. 4.4) where Tn is the current, n t h sample, SLOPE = S(T„) = M(Tn) - M(Tn.3) (eq. 4.5) ACCELERATION = A(T„) = S(T„) - S(Tn.2). (eq. 4.6) Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage Fig. 4.17: Arrival detection algorithm. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 90 System to Allow Estimation of LOP for Tourniquet-Related Usage In addition, the value utilized for comparison of acceleration is the sum of the current and previous sample acceleration. Again, this serves to attenuate the second order response from small spikes in the waveform. Q H Z (3 < > < -j ill cc V M(Tn) S(Tn) = M(Tn) -M(Tn-3) A(Tn) = S(Tn) -S(Tn-2) X(Tn) = A(Tn) + A(Tn-1) SAMPLE Fig. 4.18: The use of multi-sample differentiation to attenuate the second order response of small amplitude high frequency noise. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage This method of slope and acceleration calculation results in a delay of approximately three samples between detection of the peak acceleration point and the actual point of occurrence. 4.5.4 Noise detection In the absence of significant noise or motion artifact, the pulse arrival point is readily identified as the first sign of upturn in arterial pressure prior to the long 50-150 msec rise to the maximum point and in this case can be detected using a relatively simple algorithm as described previously. In the presence of noise or motion artifact, detection of the arrival point becomes a complex problem. The use of a matched filter to extract the underlying pulsatile signal is inappropriate and the use of cross correlation or adaptive filter techniques have been found to be ineffective in the presence of reflected waves and place a high computational demand on the microcontroller CPU as detailed previously in section 4.5.3. In finding a solution to this problem, one can look to the intended application for the PWTT system to control an adaptive surgical tourniquet system and note the following: (1) it will not adversely affect performance of an adaptive surgical tourniquet system if PWTT measurements are periodically unavailable provided that the length of the interruption is no more that 10 or 15 seconds; (2) the ECG signal will be essentially noise free, even during limb manipulation or the use of surgical tools; (3) assuming that the pulse sensor is applied properly, the signal can generally expected to be relatively noise free and any noise will be primarily due to motion artifact as the surgeon re-positions the limb or when tools such as hammers or chisels are utilized; Chapter 4: The Design and Implementation of a Prototype PWTT Measurement g2 System to Allow Estimation of LOP for Tourniquet-Related Usage (4) these periods of artifact will be intermittent, generally lasting only for a fraction of a minute or, if longer, with periodic breaks when a noise free pulse signal will be available; and (5) these noisy periods are readily distinguishable from noise free periods since there will be large amplitude signal responses during periods when no pulse is expected and characteristic features of the waveform such as the detected arrival time and the peak magnitude of the pulse signal will be observed to vary significantly from sample to sample. It then follows that a suitable method for detection of the pulse arrival will be to simply accept PWTT values for sample periods during which it is determined that no noise was present and discard PWTT values calculated for periods during which it is determined that noise was present. This will allow PWTT measurements to be obtained accurately for the majority of a surgical procedure. During periods when the surgeon is using tools or manipulating the limb, the system may reject a high portion of the transit time values, accepting only those taken during the quiet periods between the manipulations or tool work, however, this will not pose a significant problem in the effective use of an adaptive surgical tourniquet system. If no valid PWTT sample can be obtained for a lengthy period, the tourniquet cuff pressure can be increased a small safety margin to account for any increase in LOP that might occur. The controller can continue to operate in this manner increasing the pressure slightly every so often until either a "default" cuff pressure setting is reached or the signal noise ceases such that valid samples can once again be obtained. It might be possible to develop a sophisticated detection process to obtain PWTT values during these noisy periods, but the clinical suitability of this approach is questionable because Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 0 ,3 System to Allow Estimation of LOP for Tourniquet-Related Usage any values determined for the PWTT must be accurate to avoid the possibility of incorrect estimation of LOP, which might in turn result in incorrect lowering the cuff pressure and resultant leakage of blood into the surgical field. Restricting valid samples to only those as described above will achieve this end. A "pulse period" is defined as the interval between the detection of two consecutive R-waves ~ in the ECG waveform. The prototype PWTT measurement system software operates to identify noisy pulse periods by referencing a series of characteristics obtained from the most recent pulse wave to those of the previous pulse. Fig. 4.19: Pulse characteristics utilized for noise rejection routines. The characteristic features shown in Fig. 4.19 are compared to adaptive thresholds derived from the characteristics of the preceding arterial pulse to ensure that the shapes of the consecutive pulses are relatively constant. The threshold values which control the allowed Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 9,4 System to Allow Estimation of LOP for Tourniquet-Related Usage interbeat variance were determined through trial and error and minimize the possibility of a noisy signal being passed while only rejecting occasional "good" pulses. In addition to matching the shape characteristics of consecutive pulses, each pulse period is monitored for the occurrence of "extra" peaks due to noise and excessive beat-to-beat variation in time of the peak point of the pulse wave. During a clean pulse period, there should be a single large pulse due to the arterial pulse wave and the presence of additional large peaks may indicate noise or motion artifact. For the purpose of the system noise detection algorithm, noise peaks are considered any spurious peak that is greater than 50% of the magnitude of the previous pulse peak. It is also expected that the time of the maximum pressure point should be relatively slowly varying over time. The observance of a large beat-to-beat change in this interval may indicate that a sample period is corrupted by noise and therefore beat to beat changes of greater than 20 msec result in the rejection of the pulse period. Signals that are of generally poor quality will continually trigger the noise detection algorithm since their characteristics will vary randomly between pulse periods. Calculated PWTT times from pulse periods which meet the threshold criteria for the characteristics shown in Fig. 4.19 and the timing and spurious peak criteria as outlined in the above paragraph are considered as valid. The PWTT from the current pulse is then recorded in the 8 register circular "result stack" provided that that the previous pulse period was also considered as valid. In this way only PWTT values that come from signals where three or more consecutive arterial pulses "match" are used to estimate LOP values. The use of threshold values based on the previous period, valid or not, rather than from the last valid period allows the system to adapt quickly to changing waveform morphologies or rapidly changing PWTT values. Were the threshold values to be based on the last valid pulse period, it would be possible for the algorithm to continually reject samples by comparing them to a set Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 95 System to Allow Estimation of LOP for Tourniquet-Related Usage of thresholds that were either mistakenly adopted or were determined prior to a significant physiological change which resulted in a corresponding shape change in the arterial pulse wave. 4.5.5 Two-step averaging process As depicted in Fig. 4.20, PWTT values recorded in the 8 register circular result stack described in the above paragraph undergo a two-step averaging process to limit the effects of noise due to jitter in the pulse arrival detection algorithm and small beat to beat variations in PWTT due to the respiratory cycle of the subject or other physiological means. The first step of the process calculates the average PWTT from the last 8 samples. This value is then used to determine thresholds which are then utilized in the second step to allow excessively high or low values which might indicate erroneous or unstable readings to be discarded. A second average is then calculated using only samples within the threshold limits. This process is carried out after each detection of the R-wave, however, the result output register is only updated with the second step average if five or more of the eight samples fall within the thresholds determined in the first stage of the averaging process. Utilizing the noise rejection routines described in section 4.5.4 and the two-step averaging process described above assures to a clinically acceptable level that each update of the output register results from a uniform pulse stream, with little beat-to-beat variation in the wave morphology or arrival time. While this results in a slight delay in the output of PWTT values over true real time, this approach ensures that calculated values are valid trends in pulse wave velocity and not the result of waveform anomalies, motion artifact or other short term conditions which might otherwise produce corrupted results. Chapter* The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage ENTER SAMPLE IN B REGISTER CIRCULAR STACK COMPUTE AVERAGE OF B SAMPLES IN STACK AND SET UGH AND LOW THRESHOLDS BASED ON THS AVERAGE INCLUDE SAMPLE IN STEP II AVERAGE COMPUTE STEP II AVERAGE BASED ON SAMPLES WITHN THRESHOLD RANGE I UPDATE OUTPUT REGISTER TO SHOW NEW AVERAGE PWTT AS CALCULATED IN snp n Fig. 4.20: Block diagram of the two step averaging process. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 97 System to Allow Estimation of LOP for Tourniquet-Related Usage 4.5.6 User interface and output Valid updates of the second step-averaged PWTT values consist of an 8-bit value representing the number of 3 msec periods in the average PWTT. These are output to the system user via the pulse width modulation (PWM) port of the microcontroller board which converts the 8-bit number to a pulse width modulated signal. The PWM port drives a buffer and low pass filter which convert the pulse wave modulated output to a constant voltage signal ranging between 0-5v such that, PWTT m s e c = V o u t * 154. (eq. 4.7) An 8-segment LED array is utilized to provide feedback to the user as to the system operation in order to facilitate clinical testing. The LED array is configured to indicate the presence of signal noise which results in the rejection of a pulse period or indicate the number of valid samples (5-8) utilized in each update of the PWTT result output. An analog meter is utilized to provide visual indication of the relative amplitude of the arterial pulse signal in order to allow adjustment of pulse probe position and gain controls. Gain adjustment for the arterial pulse signal is made manually using a 4-element dip switch array. Circuit diagrams for the system output stages are provided in Appendix A. Power for the system is provided from two 12v rechargeable batteries connected in series to provide +13.6V and -13.6V to voltage regulators which provide constant +8 v, -8 v, and +5 v sources for the system hardware. The system ground is connected to the external ground for the other O.R. equipment through the ground wire of the ECG waveform interface. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement System to Allow Estimation of LOP for Tourniquet-Related Usage 98 4.6 Sources of Error The main sources of error in the steps leading to the output of the calculated PWTT are summarized in the following table. SOURCE T R I G G E R I N G O F R - W A V E D E T E C T O R D E T E C T I O N O F R - W A V E T R I G G E R A T HIGH S P E E D INPUT E R R O R IN L E A D I N G E D G E D E T E C T I O N E R R O R IN DIGITAL V O L T M E T E R DISPLAY ERROR +/- 5 msec (estimated from chart traces) +/- 3 msec (1 cycle) +/- 6msec (estimated from chart traces) +/-1.5 msec ( +/- .01 v dc) TOTAL +/-16msec Table 4.1: Main sources of error in measured PWTT. This error is the worst case for a single measurement from a waveform which passes the noise detection checks. Since the output of the system is actually the result of the two-step averaging routine the measurement error in the output value is reduced as a function of the number of samples utilized to calculate the average. Given that the output is an average value over 5-8 samples, the error in the value output decreases as the square root of the number of samples utilized to calculate that average such that the system error is approximately [63]: +/-16 msec = +/- 7 msec, (eq. 4.5) V5 taking the worst case of an update of the output based on a 5 sample average. Chapter 4: The Design and Implementation of a Prototype PWTT Measurement gg System to Allow Estimation of LOP for Tourniquet-Related Usage 4.7 Compliance with Electrical Safety Standards Prior to its use in clinical trials, the PWTT measurement system was reviewed by a registered Professional Engineer working independently in the Biomedical Engineering Department of the Vancouver Hospital and Health Sciences Centre to confirm that it met pertinent regulations and safety standards as required for temporary departmental approval based on those set out in the CSA Standard C22.2 Number 125, Electromedical Devices, and CSA Standard C22.2 Number 601.1 M90, Medical Electrical Equipment [59,64]. Temporary approval was obtained in order to use the system as designed for supervised data collection in Vancouver Hospital and Health Sciences Centre. Of particular concern is the potential sink current through the patient applied pulse sensor. Testing was carried out using the apparatus and test procedures as required in the above standards and the resulting sink current was found to be below the maximum allowable limit of 100 uA. Also of concern is the potential for heating of the patient applied sensor due to the power dissipation in the LED. The standards require that any patient applied part not have a surface temperature in excess of 41 °C. While no formal testing was carried out to determine the heating of the arterial pulse sensor, the sensor was applied for periods of up to several hours in extensive laboratory testing during which at no time was noticeable warming of the sensor LED noted. In addition, given that the LED is operating at 17% of the rated power dissipation, no significant heating should occur. The LED current source is fused in a manner designed to interrupt the LED current in the event that a current in excess of the design current is applied. In order to obtain CSA approval for sale in Canada, the system would need to be tested in its entirety to ensure that all requirements of CSA Standard C22.2 number 125, Electromedical Chapter 4: The Design and Implementation of a Prototype PWTT Measurement 1QQ System to Allow Estimation of LOP for Tourniquet-Related Usage Devices and CSA Standard C22.2 number 601.1 M90, Medical Electrical Equipment are met. These standards govern all aspects of the system design and construction and certification would include complete testing of heating of the patient applied sensor and physical construction of the system. Given that the PWTT measurement system is intended only to collect research data and is likely to undergo significant redesign for integration into an adaptive surgical tourniquet system detailed testing under these standards was not warranted. Chapter 5: Laboratory Evaluation of the Prototype System 1Q1 CHAPTER 5 LABORATORY EVALUATION OF THE PROTOTYPE SYSTEM 5.1 Calibration The prototype system described in Chapter 4 was developed to allow measurement of the transit time of an arterial pulse between the ejection phase of the cardiac cycle and the arrival of the resulting arterial pulse at a peripheral point on the body. Prior to undertaking extensive laboratory and clinical testing utilizing the prototype system it was necessary to determine the accuracy of the system and confirm its proper operation as designed. This section describes this performance testing and calibration of this system. 5.1.1 Calibration method In order to confirm that the detection algorithms and timing of the PWTT measurement system were functioning as designed, a multi-track chart recorder was utilized to make hard copies of the pulse waveform and ECG input. In addition, the system software was re-configured to produce flags at the high speed output of the microcontroller indicative of the operation of noise detection and arrival detection routines in the system software. These flags were displayed simultaneously on the chart recordings with the ECG and pulse signals. The software was re-configured to create positive transitions at separate pins of the high speed output port corresponding to : • each time the maximum pulse waveform acceleration register is updated; • each time the minimum pulse waveform magnitude register is updated; and • periods during which the pulse waveform is determined to be corrupted by noise. Chapter 5: Laboratory Evaluation of the Prototype System 102 The ECG signal utilized in the chart traces was obtained directly from the analog output of the ECG monitor and the R-wave detection channel was taken as the output of the analog R-wave detection circuit detailed in section 4.2. The arterial pressure waveform signal was obtained at the input to the microprocessor a/d pin. A sample chart trace is presented in Fig. 5-1 Note that the photoplethysmographic pulse waveform seen in Fig. 5.1 is inverted from the corresponding arterial pressure waveform that would be expected. This inverted response results from the fact that the amount of light reflected to the sensor photodiodes decreases as a function of blood volume in the underlying tissue. Testing was then carried out on 7 normal subjects by applying the electrodes and leads to obtain the ECG signal and applying the pulse sensor to the fingers or toes using a velcro strap or under the proximal edge of a Zimmer low pressure reusable single bladder dual port adult thigh cuff (Zimmer, Dover OH) or a Zimmer low pressure contoured reusable single bladder dual port arm cuff using the gel housing described in section 4.3.5. The chart recorder paper feed was then activated to produce a trace of the pulse sensor, ECG and software flag waveforms. PWTT values output by the system were recorded manually on the corresponding paper traces. Chapter 5: Laboratory Evaluation of the Prototype System 103 ECG ECG R-WAVE DETECTOR OUTPUT ARTERIAL PULSE WAVEFORM NEW MINIMUM UPDATE PEAK ACCELERATION UPDATE NOISE DETECTION Fig. 5.1: Sample trace from chart recorder as configured for calibration of the PWTT measurement system. Chapter 5: Laboratory Evaluation of the Prototype System -j Q 4 5.1.2 Performance of the pulse sensor The pulse sensor was found to produce strong pulse signals from the fingers and toes on the 7 subjects tested from which the software algorithms were readily able to identify the leading edges (see section 5.1.3). While the signals tended to be weaker and somewhat noisier from pulse sensors located under the tourniquet cuff, in all cases where the sensor was applied flush to the limb and centred 2 cm under the proximal edge of the tourniquet cuff the signal was found to be of sufficient quality and magnitude such that clinically accurate PWTT values could be calculated. The quality of the pulsatile signal obtained from under the tourniquet was dependent on the bladder pressure of the overlying tourniquet cuff and on the location relative to the proximal edge of the cuff. For tourniquet cuffs applied to both the upper arm and thigh, it was found that signal magnitude was drastically reduced when the sensor was located such that the centre of the sensor was greater than about 3 cm under the cuff. Poor signals could also result from the sensor being applied too close to the edge such that the sensor did not lie flush on the surface of the limb. A minimum tourniquet cuff pressure of about 30 mmHg was generally required to ensure that a usable signal could be obtained from a pulse sensor located underneath. For sensors applied greater than 3 cm under the tourniquet cuff, the signal quality and magnitude, while adequate at low pressure, decreased as a function of the pressure within the tourniquet cuff such that at normally utilized inflation pressures of 200-300 mmHg usable signals were generally unavailable. This is probably due to the occlusion of blood flow to the vascular bed underlying the sensor. For application to the upper arm, the rear of the arm was found experimentally to produce the best signal from the pulse sensor. For application to the upper thigh, the front and front medial area of the thigh produced the best signals. This may be due to better perfusion of these areas with shallow blood vessels. Chapter 5: Laboratory Evaluation of the Prototype System •] 05 Great variance between pulse signal shape and amplitude was found from case to case, even for the same anatomical site and pressure of application. This is due to the dependence of pulse shape on a wide range of physiological factors which differ with each application, subject and location on the body. Of particular interest when the sensor was applied under the tourniquet cuff was the occasional appearance of a waveform corresponding to the arterial pulse train but of reversed polarity to that normally expected. These "inverted" waveforms are otherwise periodic and include deflections which are similar to those that might be expected to result from the arrival of the arterial pressure pulse but these occur at times which do not correspond to true pulse arrival and therefore result in erroneous PWTT values being calculated. — A A — s / 1 K \ i Fig. 5.2: Sample of reversed polarity pulse waveform. It is suggested that the inverted waveforms seen in Fig. 5.2 occur when the pressure change in the bladder of the tourniquet cuff resulting from the arrival of the arterial pulse wave causes a corresponding physical shift in the pulse sensor relative to the underlying vascular bed that ejects the blood underneath, compresses the underlying tissue or otherwise creates a periodic Chapter 5: Laboratory Evaluation of the Prototype System -| QQ physical movement that increases the amount of reflected infrared light reaching the photodiodes, causing an upswing in the photoplethysmographic waveform. This is the same mechanism that serves to amplify weak pulsatile signals as noted previously; however, certain physical conditions may cause an inverted response of sufficient magnitude to obscure the true arterial pressure signal. The occurrence of these inverted waves for application of the sensor under the tourniquet cuff was found to be unpredictable in that it was not readily possible to predict exact application pressures and positions which would cause the anomaly. The shape of the inverted waveform is highly dependent on tourniquet cuff pressure and in situations where inverted pulses were found to occur, the pulse shape generally changed markedly with tourniquet pressure, often returning to a "normal" pulse shape as the tourniquet pressure was reduced or increased beyond a certain arbitrary level. Inverted signals were also found to occur more frequently in thigh cuff applications than in upper arm or forearm applications for the 7 normal subjects tested in the laboratory. Inverted waveforms were not observed for application of the sensor to the fingers or toes for the 7 normal subjects tested in the laboratory. During laboratory evaluation it was possible to identify the occurrence of inverted waveforms and adjust the sensor position slightly so as to obtain a normal signal in ail 7 subjects. However, for use in a fully automated system it will be necessary to either modify the method of application of the pulse sensor to prevent the occurrence of inverted waveforms or implement software routines which detect these waveforms and then request that the user re-position the sensor. Further investigation of the phenomenon of pulse inversion was considered to be beyond the scope of this research. Chapter 5: Laboratory Evaluation of the Prototype System *\QJ 5.1.3 Performance of pulse arrival detection algorithms In analysis of the chart paper traces, the arrival point of the pulse detected by the system is identified as the last maximum "downward" acceleration register update prior to the last minimum pulse magnitude update in each cardiac cycle. This is illustrated in Fig. 5.3. Analysis of sample traces obtained from the 7 normal subjects for a variety of sensor locations confirmed that the detection algorithm identifies a consistent point on the leading edge of the pulse wave. Representative sample traces are included in Appendix C. The pulse arrival point as detected by the system was typically about 12 msec later than the first sign of slope upturn. This delay results from the slope and acceleration algorithms which utilize the current sample and that two and three intervals back to calculate the first and second derivatives and the use of the summation of second derivative values for peak acceleration comparison purposes as detailed previously in section 4.5.3. This constant 12 msec delay is not of concern since it does not affect the measurement of changes in PWTT. It was determined through analysis of chart traces from 32 separate probe applications over 7 subjects that for arterial pulse waveforms of 1-3 V in magnitude the exact point of arrival determined by the system varies randomly from beat-to-beat on the order of +/-6 msec as a result of the cumulative effects of resolution errors in the leading edge detection routines. This analysis comprised approximately 320 cardiac cycles. Chapter 5: Laboratory Evaluation of the Prototype System 108 Notes: [1] Signal from analog output of ECG. [2] Signal from output of analog R-wave detector. Positive transition corresponds to triggering of the threshold circuit to indicate the occurrence of the R-wave. Pulse duration is approximately 150 msec. Software routine checks every 3 msec for this transition which signals the end of the last cycle and the trigger point for calculating the PWTT for the next cycle. [3] The pulse signal from the arterial pulse probe as seen at the input to the a/d converter. [4] "NEW LOW" indicator. This is reset after each R-wave is detected and makes a positive transition each time a new minimum is detected in the pulse waveform. This function is not activated until approximately 50 msec after each R-wave is detected. At each new low, the time of the last peak downward acceleration is recorded as the potential pulse arrival point. [5] "MAX ACCELERATION" indicator. This is reset to zero after each positive or zero slope and the trace waveform is configured such that it makes a positive transition each time a new maximum downward acceleration is detected. THE ARRIVAL OF THE LEADING EDGE OF THE PULSE AS DETECTED BY THE SYSTEM SOFTWARE IS THEREFOR INDICATED AS THE LAST POSITIVE TRANSITION OF THIS WAVEFORM PRIOR TO THE LAST "NEW LOW" UPDATE FOR EACH CYCLE. [6] "NOISE INDICATOR". Makes a positive transition whenever a sample is rejected by the noise routines. Not reset to zero until two consecutive sample periods are passed by the noise detection routines. Fig. 5.3: Determination of pulse arrival detection point. Chapter 5: Laboratory Evaluation of the Prototype System 1 gg The accuracy of the arrival detection algorithms was found to vary slightly with the amplitude of the pulse signal at the A/D inputs with very large (5 V p-p) or small ( < 1 V p-p) waveform amplitudes introducing significantly more beat-to-beat jitter than for midrange waveforms of 1-3 V p-p in amplitude. For the purposes of further testing the gain was adjusted manually to ensure that the signal amplitude was within this range. It is expected that in clinical use an automatic gain control circuit would perform this function. 5.1.4 Performance of noise detection algorithms The noise detection algorithms were tested by applying the PWTT measurement system to the subject and then simulating the limb manipulations that would normally be performed during a surgical procedure. It was not possible to simulate noise that might result from the use of pneumatic tools or the hammering and chiseling of bone. The noise detection routines were found to be functioning as intended, identifying any sample cycle which did not match the basic characteristics of the previous period (see Fig. 5.4). The threshold constraints tended to be more stringent than actually required, resulting in the discarding of occasional "good" pulse periods where although the signal was uncorrupted by noise or artifact, relatively large beat-to-beat fluctuations in pulse characteristics triggered the noise detection algorithms. As designed (see section 4.5.4), after detecting a noisy period the system did not consider the samples to be once again noise-free until two consecutive cycles matching the threshold requirements were recorded. When the pulse sensor was applied poorly such that a weak, noisy or highly variable signal was obtained, the noise detection flag was found to remain on continuously until the sensor was re-positioned or adjusted such that a clean signal was obtained. Chapter 5: Laboratory Evaluation of the Prototype System 110 Fig. 5.4: Performance of noise detection routines. Chapter 5: Laboratory Evaluation of the Prototype System -\ -\ -| 5.1.5 Accuracy of PWTT values calculated by the system Given that the leading edge detection algorithms were found to be functioning as intended subject to the small delay described previously and that the R-wave detection circuit was found to trigger at a consistent point representing the midpoint of the Q-R transition in the ECG waveform as detailed in section 4.2, the PWTT values calculated by the system were checked for accuracy against values calculated by hand from ECG and arterial pulse waveforms on the associated chart traces as described below. The system was applied to measure the PWTT to the index finger of seated subjects. Subjects were asked to relax and remain still in order to produce stable PWTT results updated on each cardiac cycle. The paper feed on the chart recorder was then allowed to run continuously at 25 msec/div for a period ranging between 5-15 seconds. Each update of the PWTT output was recorded at the corresponding point on the paper trace each time the user interface LED display indicated that the second average routine had produced an output value based on an 8 sample average. Given that no noise was detected during the previous 8 cardiac cycles, the PWTT values calculated and output by the system are the average of the individual PWTT values from the 8 sample periods immediately preceding period at which the output is updated. See section 4.5.5 for details of the two-step averaging process utilized to generate the measurement output. Traces were obtained from 7 normal subjects and the corresponding PWTT values determined by hand by measuring the number of time scaled divisions between the triggering of the R-wave circuit detector and the detection point of the leading edge of the corresponding arterial pulse wave. Given a hand measurement error of about +/- 2 msec on the start and arrival points of the pulse wave on the chart traces, the error in each PWTT calculated by Chapter 5: Laboratory Evaluation of the Prototype System -| -| 2 hand is expected to be approximately +/- 4msec. Averaging the 8 samples immediately previous to each PWTT output value as described above produced values that in ail cases corresponded within error to the PWTT values calculated by the microprocessor system. It was therefore determined that the PWTT measurement system was suitable for calculating PWTT values with sufficient accuracy to facilitate clinical testing to determine the suitability for using measured PWTT to estimate changes in LOP. 5.2 Laboratory Testing 5.2.1 Purpose The purpose of the initial laboratory testing was to verify in practice the inverse relationship between PWTT and arterial BP by obtaining representative measurements over a range of blood pressures from a small sample of subjects. Given that the relationship was found to hold true it was then desired to quantify the magnitude of the PWTT shifts for known BP variations and ascertain the validity of the use of generic slope coefficients to estimate changes in LOP from measured PWTT. In order to perform this testing, it was necessary to obtain a baseline (PWTT.BP) pair relating the subjects normal resting BP to a measured PWTT. It was then desired to cause shifts in BP which could then be related to new PWTT values measured at these new pressure levels. In order to minimize measurement errors in both PWTT and BP measurements it was desired that shifts in BP be as large as possible and relatively constant with regard to the length of time required to make measurements. Chapter 5: Laboratory Evaluation of the Prototype System -| -j 3 5.2.2 Causing blood pressure shifts As detailed previously, experiments involving elevation of the legs from the prone position in an attempt to increase blood pressure in the upper body actually resulted in a small decrease in blood pressure as the body acted to maintain its desired BP level. This is a result of the baroreceptor reflex which serves to regulate BP as the body changes position [25]. Baroreceptors are stretch receptors located in the aortic arch and the carotid sinus (the confluence of the large arteries feeding the head and brain areas) which serve to stimulate the heart and circulatory system to increase or decrease blood pressure through control of cardiac output and peripheral resistance. When a person goes from a lying to a standing position, there is a shift of 500-700 ml of blood from the veins of the thoracic cavity to veins in the lower extremities which extend to contain the extra volume of blood. This pooling of blood reduces the venous return and the cardiac output [25]. The resulting fall in blood pressure is almost immediately compensated by the baroreceptor reflex which ensures that the blood supply to the brain remains at a relatively constant level. Any attempt to change the blood pressure in the head or torso area by limb elevation will be counteracted by the body's natural pressure regulation system. The use of gravity inversion devices, by which a person can be hung upside down have been found to cause a significant elevation in both systolic and diastolic pressure. In one study of 50 normotensive men and women, inversion caused systolic blood pressure to rise from an average of 114 mmHg to nearly 140 mmHg, whereas diastolic pressure increased from 76 to 91 mmHg [24]. Such pressure rises involve setting up a large pressure gradient along the measurement paths due to the effect of the standing column of blood in the legs and torso and therefore inversion is not considered a valid method to obtain data relating to a subject in a prone position as seen in the operating room. Chapter 5: Laboratory Evaluation of the Prototype System ^ -j 4 The use of the valsalva maneuver, where exhalation is attempted against a closed glottis, causes increased thoracic pressure which results in the compression of the vena cava (the large veins which flow into the heart) which interferes with the return of blood to the right atrium. This results in a lowering of the cardiac output with a corresponding drop in blood pressure that is quickly compensated for by the baroreceptors which cause an increase in peripheral resistance. When the glottis is finally opened and the air exhaled, cardiac output quickly returns to normal while peripheral resistance remains elevated resulting in a sudden increase in blood pressure [24,25]. Unfortunately, the duration of the events resulting from the valsalva maneuver ( approximately one minute ) is too short to obtain accurate BP data using existing non-invasive devices such as an automatic oscillometric monitor or by using a cuff and K-sound method. The use of an arterial catheter which would provide accurate continuous results is not acceptable for these trials because of the invasive nature. Weltman et al. [40] found that the typical pattern of PWV as recorded from laboratory test subject performing the valsalva maneuver correlated well with typical arterial pressure responses as recorded by others. Unfortunately the inability to obtain an accurate correlation between BP and PWTT due to the short duration of the events makes valsalva maneuver testing of little value for the purposes of this project. The remaining options available to create the type of short term shift required for testing involve either "stressing" the subject or the use of drugs. The use of drugs is obviously not suitable for use in initial laboratory testing, but, drugs are a factor in the BP shifts seen during surgery and therefore shall be an effector in the clinical testing covered in the next chapter. The remaining option is the use of "stress" testing such as cold pressor or exercise based tests. Chapter 5: Laboratory Evaluation of the Prototype System -j^g Cold pressor testing relies on the immersion of the hand in cold water to trigger a response to external cold stimulus which results in a marked shift in blood pressure. Wood et al [65] and Harlan et al [66] found that it is possible to raise the blood pressure of subjects by having them immerse their hand to above the wrist in 4°C ice water for a period of one minute. Studies have shown the SBP shift mean to be on the order of 17 mmHg in normotensive subjects. This shift is expected to peak about 30 seconds after immersion and return to baseline levels within 1 minute of removal from the icebath [67]. While this shift is relatively short, it was hoped that it might be possible to obtain satisfactory PWTT vs. BP measurements using the method. Unfortunately in practice it was found that the BP shifts were often unpredictable, small and hard to measure due to their transience. As such, cold pressor testing was aborted after 2 subjects with no usable data obtained. 5.2.3 Exercise-based stress testing The method of stress testing most frequently used clinically involves the use of exercise. In rhythmic muscular activity such as jogging or bicycling, the dilation of the blood vessels in the working muscles enhances the flow of blood through a large portion of the peripheral vasculature. The alternate contraction and relaxation of the muscles also provides a significant pumping force to propel blood through the vascular circuit. The increased flow during moderate rhythmic exercise causes systolic blood pressure to rise rapidly in the first few minutes of exercise leveling off at 140-160 mmHg [24]. Elaborate stress tests utilizing a variable speed treadmill are routinely used in the detailed assessment of cardiac patients. For the purpose of the testing carried out for this project, subjects were required to pedal a stationary bicycle for approximately 2 minutes in order to elevate their BP. BP and PWTT Measurements were then taken at regular intervals as their BP returned to pre-exercise Chapter 5: Laboratory Evaluation of the Prototype System -| 1 g baseline levels. BP measurement was made using a standard BP cuff connected to a digital manometer applied at the upper arm and a Doppler flowmeter applied over the radial artery at the wrist. The BP cuff inflation pressure at which the pulsatile flow of blood disappeared at the radial artery was considered to provide an approximate estimate of SBP. This method of measurement was chosen over the use of an automated oscillometric BP measurement system which tends to be quite slow in taking readings and is particularly unsuited for use when BP is changing rapidly. Only SBP was measured since this allowed rapid measurements to be taken as opposed to measuring both SBP and DBP. Also, since the subject's BP was to be dropping over the measurement period, estimates of SBP and DBP would not be properly "matched" since the two parameters are determined at different times. In order to maximize the accuracy of the calculated transit times, measurements were made applying the pulse sensor at points with good capillary perfusion, namely the fingers and toes. These points tend to produce strong signals that are relatively noise free and were thus considered ideal for this preliminary testing. These points also have the benefit of providing the longest possible arterial path therefore reducing the effects of timing errors on the accuracy of the measured PWTT values. A summary of the experimental procedures and results for the exercise based testing is included in Appendix D. 5.2.4 Results of exercise-based tests Exercise-based testing was carried out on 4 subjects as follow: B1 - male, 25 years old, resting BP 110/70, muscular limb physiology; B2 - male, 24 years old, resting BP 130/70, normal limb physiology; Chapter 5: Laboratory Evaluation of the Prototype System <| -] j B3 - female, 24 years old, resting BP 120/80, normal limb physiology; B4 - male, 30 years old, resting BP 120/80, muscular limb physiology. While this small sample population is not representative of the general population it was deemed acceptable for these exercise-based tests. In performing these tests, several complications were encountered which introduced significant error to the results obtained. Firstly, difficulty was encountered in accurately tracking the rapidly falling BP of the subjects in the initial post-exercise period. After deflation of the BP cuff to detect the first sign of pulsatile flow in the radial artery, indicating SBP, this flow would immediately cease as the subject's SBP quickly dropped below the value just measured. Further reduction of the cuff pressure to restore the pulsatile flow in the radial artery would then result in a SBP reading significantly lower than that obtained initially. The rapidly falling SBP during the initial period after the cessation of the exercise routine coupled with the 8 sample averaging algorithm utilized in the PWTT calculation, prevented the accurate pairing of SBP values with corresponding PWTT values. The resulting error in the BP measurements as they relate to PWTT is estimated at +/- 10 mmHg. This error is most evident during the initial minute of the post-exercise period during which the BP and PWTT are most rapidly changing and decreases markedly as BP returns to near resting levels. It was also noted that PWTT values in the period following exercise never returned to the values measured prior to undertaking the exercise routine. The PWTT values recorded for post-exercise resting SBP were always slightly shorter (10-30 msec) than those measured for equivalent SBP values recorded prior to commencing exercise. This effect was noted to be more prevalent in the legs than the arms. This hysteresis appears to stem from the short term change in muscle tonus caused by the exercise routine. It can be expected that the Chapter 5: Laboratory Evaluation of the Prototype System -| -| g compliance of the arterial path would be much higher in the pre-exercise case when the muscles are relaxed than in the post exercise period when the muscles are tensed and muscle tonus is increased. The decreased compliance of the arterial path during this period could explain a shift in the PWTT from the pre-exercise level. Based on this hypothesis, pre-exercise PWTT values are not utilized in the following data analysis. Only post exercise measurements during which the tissue compliance can be expected to remain relatively constant are considered, although there is still a gradual relaxing of the muscle tissue over the course of the measurements. In spite of the problems encountered, the results of the trials did show a marked dependence of PWTT on subject SBP in the four subjects tested. A least-squared sum of errors analysis was utilized to fit a linear regression model to the data obtained for the arm and leg tests performed on each of the 4 subjects. In the case of the arm trials, the slope coefficients determined in each case were found to be relatively constant between the four subjects, with a mean value of 0.63 and a mean coefficient of correlation of 0.89. In testing performed on legs, three subjects provided highly similar slopes coefficients of an average value 0.37 while one was fit to a slope of 0.84. It should be noted that the results obtained from this subject were found to be particularly erratic due to rapidly fluctuating blood pressure and mild hypertension. The mean coefficient of correlation for these four leg cases was 0.91. In spite of the errors associated with these results, a good relationship between PWTT and SBP was found to be evident. In addition, the general agreement of the slope coefficient values determined for the least-squared error linear regression models indicated a possibility for determining a set of generic slope coefficients for various sensor locations. Chapter 5: Laboratory Evaluation of the Prototype System -| -| g In order to verify the results obtained in these preliminary trials and obtain slope information of sufficient accuracy to facilitate the development of a general algorithm for use in the control of an adaptive tourniquet, it was necessary to adapt the experiment so as to either avoid or compensate for the problems caused by rapidly changing subject BP and hysteresis due to muscle tonus. The obvious method to overcome these problems was to perform testing on subjects who are under the influence of anesthetic in the operating room, who typically undergo gradual SBP shifts on the order of 10-50 mmHg. Data obtained from such subjects would allow the accurate pairing of PWTT values measured using the prototype system with BP values as measured by the anesthesiologist. As the patient is not conscious during the measurement process it is expected that changes in muscle tonus or spontaneous shifts due to subject awareness of the measurement process will not be significant. In addition, it was necessary to obtain representative data from a wide range of subjects of differing age and health. This clinical testing is the subject of the following chapter. Chapter 6: Evaluation of the Prototype System -|20 CHAPTER 6 INITIAL CLINICAL EVALUATION OF THE PROTOTYPE SYSTEM 6.1 Preparations for Clinical Trials 6.1.1 Obtaining approval for study Before any new medical device can be evaluated on patients undergoing treatment at one of the university hospitals, an experimental protocol must be developed and a proposal submitted and approved by the Clinical Screening Committee of the Faculty of Medicine at the University of British Columbia. The screening committee consists mostly of physicians who examine primarily the ethical aspects of the proposed experiment or medical device evaluation. Factors considered are the risks and benefits of the experiment to the patient and whether or not the patient will receive the quality of care they would receive under normal conditions. Following approval on ethical grounds, additional approval must be received from the Research Committee of the hospital where the experiments are to be conducted, in this case the Vancouver Hospital and Health Sciences Centre at both the 12th avenue site and the U.B.C. site. In this review, the concerns are more administrative and technical focusing on how the experiment may impact patient care, clinical staff and hospital resources, and whether the equipment used meets hospital standards. In the case of these trials, approval was obtained as a portion of a larger project entitled "Development and Evaluation of Applied Pressure Transducers for Biomedical Applications" headed by Robert W. McGraw, Professor and Head of the U.B.C. Department of Orthopedics and James A. McEwen, an Adjunct Professor with the U.B.C. Electrical Engineering Chapter 6: Evaluation of the Prototype System -|21 Department. The purpose of the overall project is to improve the safety of medical devices which intentionally apply pressure to a portion of the human body. This is being done by developing applied pressure transducers, by integrating these transducers into selected medical devices and by clinically evaluating these transducers and integrated medical devices. The investigation of the use of PWTT as an indicator of blood pressure changes is considered as part of the development process for an adaptive tourniquet with provision for applied pressure sensing. Copies of the approval documents are included in Appendix E. Two orthopedic surgeons were selected for participation in these trials, Dr. McGraw and Dr. Brian Day, and both were consulted about and approved of the experimental procedures. Dr. Day specializes in arthroscopic surgery of the knee and shoulder and Dr. McGraw in knee and hip arthroplasty. Prior to its use in these trials, the prototype PWTT measurement system was evaluated independently by Professional Engineer employed in the Biomedical Engineering Department of the Vancouver Hospital and Health Sciences Centre to determine weather it met pertinent regulations and safety standards as required for approval for temporary use in the hospital based on requirements set out in the CSA Standard C22.2 number 125, Electromedical Devices and CSA Standard C22.2 number 601.1 M90, Medical Electrical Equipment [59,64]. Temporary approval was obtained which allowed use of the system as designed for supervised data collection in Vancouver Hospital. Chapter 6: Evaluation of the Prototype System -|22 6.1.2 Test subjects Subjects chosen for the study were a random sampling of patients of Dr. McGraw and Dr. Day who were scheduled for orthopedic surgery. In the initial study, a total of 16 subjects were tested as detailed in Table 6.1. SUBJECT ID A G E SEX ANESTHETIC PROCEDURE BD1 35 F GENERAL ARTHROSCOPIC KNEE SURGERY BD2 29 M GENERAL BONE STAPLE REMOVAL BD3 72 M LOCAL ARTHROSCOPIC KNEE SURGERY BD4 24 F GENERAL ARTHROSCOPIC KNEE SURGERY BD5 34 M GENERAL ARTHROSCOPIC KNEE SURGERY BD6 28 M GENERAL ARTHROSCOPIC KNEE SURGERY BD7 27 M GENERAL ARTHROSCOPIC KNEE SURGERY BD8 56 M GENERAL ARTHROSCOPIC KNEE SURGERY BD9 27 M GENERAL ARTHROSCOPIC KNEE SURGERY BD10 61 M GENERAL ARTHROSCOPIC KNEE SURGERY BD11 34 F GENERAL ARTHROSCOPIC KNEE SURGERY BD12 29 M LOCAL ARTHROSCOPIC KNEE SURGERY RM1 66 F LOCAL REPLACEMENT OF KNEE PROSTHESIS RM2 70 F LOCAL HIP ARTHROPLASTY RM3 73 M LOCAL KNEE ARTHROPLASTY RM4 77 M LOCAL HIP ARTHROPLASTY Table. 6.1: Sample subjects for initial clinical testing. An additional 8 subjects were tested using procedures developed based on the results of testing on these initial 16 subjects. This is detailed in Chapter 7. Test subjects in this study are not identified by name in order to protect patient confidentiality. This sample population represents a good cross section of the type of surgeries during which an adaptive tourniquet would typically be used. Two hip surgeries are included in this study Chapter 6: Evaluation of the Prototype System -\ 23 which do not require the use of a tourniquet but they did allow collection of data useful for this investigation. All subjects in this study had the arterial pulse sensor applied at the finger tip. 6.2 Experimental Protocol Prior to admission of the patient to the operating room, the measurement system was connected to the operating room ECG monitor and tested to ensure that it was functioning properly. After each subject was admitted to the operating room and positioned on the operating table, the arterial pulse sensor was then applied to the finger at the earliest opportunity and the gain adjusted to obtain an adequate arterial pulse signal from the sensor. Unless required due to a large change in signal amplitude, this gain then remained fixed for the duration of the measurements. The system was then checked to confirm proper operation and that consistent PWTT values were being measured prior to beginning to record data. Each BP as determined periodically by the operating room BP monitor (an oscillometric type device with the BP cuff applied to the upper arm) was then recorded in conjunction with the associated PWTT value as determined by the prototype measurement system. Typically the operating room patient monitor is set by the anesthetist to take BP readings at 4 minute intervals. Where the PWTT was found to change significantly over the duration of the oscillometric BP cuff measurement process, PWTT values were recorded at both the initial inflation of the BP cuff and at the point at which the BP measurement was completed. The use of an oscillometric BP reading for comparison purposes is not ideal due to inherent problems with the technique as discussed in Chapter 2; however, no other measurement Chapter 6: Evaluation of the Prototype System -|24 method is generally available for use in orthopedic surgeries. Testing on patients with direct invasive BP monitoring through an arterial catheter such as those undergoing some cardiac surgeries might prove suitable but access to such procedures was not available for this project and the resulting data might not have general clinical application in any event. A representative sample of the data collected is included in Appendix E. 6.3 Performance of the System in Obtaining PWTT Values The pulse sensor was found to perform well in the majority of the cases tested, producing a strong, stable pulsatile signal representative of the arterial pulse wave as viewed simultaneously on the analog display of the pulse oximeter. A pulse oximeter is a device utilized to measure blood oxygen levels that is commonly utilized by the anesthesiologist to monitor a patient during surgery. A pulse oximeter uses a photoelectric sensor generally applied to the finger and provides a photoplethysmographic display in addition to measuring blood oxygen levels. In some cases where the patient was particularly poorly perfused or cold, the pulse signal was weak and subject to frequent rejection by the error detection routines; however, in all cases the signal was adequate for obtaining PWTT values. The noise rejection and averaging routines functioned to produce a stable measurement output that was found to be resistant to motion artifact and other noise. By observation it was determined that no erroneous PWTT values were generated as a result of motion artifact. The use of electrosurgical equipment to cauterize and occlude arteries in the hip replacement surgeries was found to create large pulses of noise in the measured pulse waveform. However, these noise pulses were readily detected and rejected by the software noise detection routines. Chapter 6: Evaluation of the Prototype System 125 As expected, the use of the oscillometric BP cuff was somewhat prone to error as a method for calibration of the system periodically producing erroneous estimates. It is known that the oscillometric BP monitors utilized in the operating room will generate artificially high values in the measurement taken immediately following a measurement cycle which is unable to obtain a set of BP values due to a noisy signal, motion artifact of other such corruption. Therefore occasional "misses" of artificially high or low values as determined in consultation with the anesthesiologist in each case were discarded as data points. One case was aborted due to low battery power in the system and two cases showed insufficient change in BP over the duration of the procedure to warrant inclusion in the data analysis. 6.4 Results of Clinical Trials for Ascertaining the Validity of the Developed Adaptive Tourniquet Pressure Algorithm 6.4.1 Short cases on relatively young patients During shorter procedures (of 1.5 hours or less), performed on relatively young subjects (less than 60 years of age) a good correlation was found between SBP and PWTT. The data from these subjects is plotted in Figs. 6.1-6.10 on the following pages. Measured PWTT values are subject to an error of approximately +/- 7msec, measured BP values approximately +/-5mmHg. A least squared error linear approximation fitting the data points is superimposed on each set of results. The mean correlation of these cases with their respective least squares fit lines is 0.86. The slope values determined from the linear models ranged from 0.24 - 0.97. Chapter 6: Evaluation of the Prototype System 126 152 147 142 137 132 127 122 117 112 107 BD1 * measured data ° 373-0.97"PWTT 240 250 260 PWTT (msec) 112 110 108 & 106 E E " 102 gj ' 0 0 104 BD2 measured data D 241-0.44 "PWTT 310 320 PWTT (msec) Fig. 6.1: PWTT vs. BP-BD1 Fig. 6.2: PWTT vs. BP - BD2 i E E CL CQ (0 130 128 126 124 BD3 • measured data ° 188-0.27'PWTT 240 260 PWTT (msec) a too cn tn BD4 measured data ° 201-0.33 "PWTT 300 320 340 PWTT (msec) Fig. 6.3: PWTT vs. BP - BD3 Fig. 6.4: PWTT vs. BP - BD4 Chapter 6: Evaluation of the Prototype System 127 150 145 140 o 135 X E 130 £ 125 0L 120 CQ CO 115 110 105 100 BD5 • measured data ° 277-0.6"PWTT 260 280 PWTT (msec) •g 130 E CL 120 CQ co BD7 JtDfl , measured values ° 235-0.46 "PWTT 250 260 270 280 290 300 310 320 PWTT (msec) Fig. 6.5: PWTT vs. BP - BD5 Fig. 6.6: PWTT vs. BP - BD7 112 110 _ 108 co I E 106 E ~— 104 o. CQ CO 102 100 08 BD8 measured data ° 220-0.36"PWTT 310 320 330 PWTT (msec) MO 130 "cn X 120 E 110 0. 03 CO 100 90 80 BD9 measured data 234-0.4"PWTT "2b • 280 330 PWTT (msec) Fig. 6.7: PWTT vs. BP - BD8 Fig. 6.8: PWTT vs. BP - BD9 Chapter 6: Evaluation of the Prototype System 128 130 125 120 "5 X 115 E Ul) 110 BP 105 tn 100 95 90 BD10 B measured data ° 276-0.6 "PWTT CD 280 300 PWTT (msec) BD11 • m e a s u r e d d a t a n 176 -110 0.24-PWTT 108 • mmHg) 108 lOd 102 co. tn 100 98 <1« • 280 290 300 310 320 PWTT (msec) Fig. 6.9: PWTT vs. BP - BD10 Fig. 6.10: PWTT vs. BP - BD11 Correlation between MBP and PWTT was found to be poor with an average coefficient of correlation of 0.58 for cases BD1-BD12. In each individual case, the correlation was found to be lower with MBP than with SBP. This may be because PWV is more dependent on the peak pressure during the cardiac cycle than the mean pressure. Using the first data point obtained in each case to determine APWTT vs. ASBP and plotting these pairs for cases BD1-BD12 on the same graph the APWTT vs. ASBP becomes apparent in Fig. 6.11. Chapter 6: Evaluation of the Prototype System 1 2 9 Fig. 6.11: APWTT vs. ASBP for case BD1-BD12. Using least-squared error analysis to fit the entire data set from all subjects yields the relation: ASBP = 1.58 - 0.56 * APWTT (eq. 6.1) Using this slope (0.56), ASBP values were calculated for each measured APWTT as determined in comparison to the first PWTT value recorded for each trial. These were then plotted against the measured ASBP values and the error determined. Plotting the absolute value of the errors as a function of the magnitude of SBP change from the initial reading Fig. 6.12 is obtained. Chapter 6: Evaluation of the Prototype System 130 40 j 30 -20 --20 --30 -• -40 -• CHANGE IN SBP (mmHg) Fig. 6.12: Error in estimated SBP as a function of ASBP. 6.4.2 Long cases under local anesthetic Three trials were carried out on subjects under local (epidural) anesthetic. Two of the three cases provided usable results. One of the three cases provided no usable data due to a noisy ECG signal which inhibited system operation. One additional case was aborted due to failure of the battery on the system. In the two cases for which usable results were obtained, RM2 and RM3, the patients were elderly (>70 years) and were considered to have sufficient risk factors to make the use of a general anesthetic potentially life-threatening. These patients were undergoing arthroplastic procedures rather than the less traumatic arthroscopy. It was found that these patients Chapter 6: Evaluation of the Prototype System -| 31 tended to have erratic heart beats of rapidly varying amplitude and interbeat interval. By monitoring the photoplethysmographic display from the pulse oximeter it was possible in most cases to see noticeable beat to beat changes in the morphology of the arterial pulse waveforms as detected at the finger tip. Both patients also had a tendency to get "cold" during the surgical procedure with a resulting decrease in peripheral circulation which at some times resulted in the virtual disappearance of the pulse signal at the probe location. Discussion with the attending anesthesiologist indicated that this was typical of elderly patients. This pulse "inconsistency" tended to cause problems for the arrival-picking algorithm by continually triggering the noise detection routines which are designed to only "pass" pulse signals of relatively constant shape and timing characteristics. As such the PWTT averaging routine was relatively slow to update the output due to the high percentage of rejected pulse periods and therefore was not able to effectively track rapid changes in pulse periods in these patients. In both patients RM2 and RM3, it was found that PWTT showed poor correlation with arterial blood pressure. The data collected in these cases is presented in Figs. 6.13 - 6.14 on the following pages. Although the error associated with poor pulse signal quality and stability was significant, this cannot fully explain the lack of correlation between the PWTT and SBP for the gross changes in BP observed. It is hypothesized that this lack of correlation is a result of two factors explained below. Chapter 6: Evaluation of the Prototype System 132 E E 95 | 9 0 RM2 300 320 PWTT (msecl 200 190 -» 180 01 X E 170 E ~ 160 ca <n 150 140 130 RM3 Oo " measured data ° 470-1.3"PWTT 230 240 250 PWTT (msec) Fig. 6.13: PWTT vs. BP - RM2. Fig. 6.14: PWTT vs. BP - RM3. Firstly, the administration of a variety of drugs during the surgery which cause changes to the cardiovascular system by altering arterial compliance may have caused intraoperative changes in the PWTT/BP relationship over the selected arterial path. The administration of such drugs may make invalid the assumption that the arterial path remains physically constant except for pressure which is used in the derivation of equation 3.9. While the use of drugs does not appear to be a significant factor in the case that general anesthesia is utilized in shorter arthroscopic procedures the greater frequency and variety of drug administration that is required to maintain the "stability" of a locally anesthetized elderly patient may prohibit a stable PWTT/BP relationship. Vasodilator agents appear to have different effects on different sized conduit arteries with the least effect on the aorta because it is naturally relatively elastic and the most at the tiny, normally rigid peripheral arteries. These peripheral arteries can display a marked constriction and can even occlude completely from localized spasm [68]. Chapter 6: Evaluation of the Prototype System 133 Commonly used vasodilator drugs such as nitrates, calcium antagonists and ACE inhibitors can have profound effects on the peripheral arteries even doubling their compliance. While experiments have shown that PWV in the aorta is not affected by the administration of vasodilators due to the already elastic nature of the arteries at that location, PWTT measurements taken over a longer arterial path to the periphery are likely to be affected by such agents independent of blood pressure due to the resulting changes in the compliance of the arterial walls towards the periphery. Secondly, it is known that in patients over the age of 70, the elasticity of their arteries is greatly diminished as the proportion of elastic fibers in the arterial walls compared to rigid fibers becomes very small. This is particularly true of the large arteries in the thoracic region. As this elasticity decreases beyond a certain point the arterial vessel may better be approximated by an inextensible tube than an elastic one. Equation 3.3 describes a wave propagating along an elastic wall. When the wall is no longer compliant relative to the pressure applied from within, the PWTT/BP relationship breaks down and equation 3.3 can no longer be utilized to accurately model the system. Estimation of BP based on measured PWTT may not be possible in these cases. This hypothesis explains a series of data obtained from a 73 year old male (RM3) where the subject's blood pressure increased steadily to over 180 mmHg (confirmed by the eventual leakage of blood under the tourniquet which was set at 300 mmHg) while the PWTT measured stayed essentially constant relative to the values seen for SBP < 140 mmHg. This clearly represented a situation where PWTT was independent of changes in SBP. In such subjects, the administration of vasodilating drugs would have an even larger effect on the PWTT/BP relationship than in younger patients since Chapter 6: Evaluation of the Prototype System - J34 dilation could be expected in not only the normally rigid peripheral arteries but also in the aorta and larger vessels. In conclusion, it is suggested that the use of measured PWTT to estimate BP changes and thereby LOP in elderly subjects, particularly when undergoing operations requiring the administration of various vasoconstrictor and vasodilator agents is highly subject to error. 6.4.3 Transient blood pressure changes The system seemed to be able to detect rapid, transient changes in BP that either happened during the interval between oscillometric cuff measurements or were missed by the oscillometric BP measurement device because of their transience. Sudden peaks were often detected in BP levels by the PWTT system in situations where the surgeon was performing gross manipulations of the limb or during periods of rapid increase in heart rate or the patient beginning to wake up during the procedure. These PWTT changes were stable and not a result of noise or artifact and coincided with physical situations in which it would be expected that the BP would suddenly increase. Unfortunately since these usually occurred in the period between oscillometric BP measurements, BP readings were unavailable to confirm these observations. 6.4.4 Other findings The assumption of a stable clean ECG signal always being available from the operating room ECG monitor to utilize for triggering purposes was found to be incorrect. In one test case (RM4) the ECG signal had sufficient baseline wander and noise content to prevent reliable triggering of the R-wave detection circuit. In this particular case, the corruption of the signal Chapter 6: Evaluation of the Prototype System 3^5 was a result of poor electrode application on the part of the pre-operative preparation nurses as well as an abnormally high amount of body hair on the chest of the individual. Discussion with the attending anesthesiologist confirmed that such situations do occur occasionally, and that in most of cases where obtaining a clean ECG signal trace is not of vital importance the anesthesiologist will not make a great effort to correct the problem. In test case RM4, surgical tape was applied over the electrodes in an attempt to secure them more tightly to the chest. This resulted in some improvement of the signal but not enough to allow the ECG signal to be used to obtain PWTT data. Chapter 7: Improvements for Application as a Clinical Tool CHAPTER 7 IMPROVEMENTS FOR APPLICATION AS A CLINICAL TOOL 136 7.1 Conclusions from Laboratory and Clinical Evaluations In the analysis of the data obtained from the clinical testing described in Chapter 6, several conclusions can be made. Firstly, in cases involving relatively young subjects, not under the influence of drugs which affect arterial compliance, a good relationship generally existed between SBP and PWTT. In these cases monitoring PWTT provided a useful means for estimating changes in LOP, particularly for short and transient changes which typically occur during limb manipulation. Secondly, for the cases in which the SBP - PWTT correlation was strong it was found that the slope and intercept of the least-squared error linear models tended to vary from subject to subject. In order to ensure maximum accuracy in estimating LOP changes it was therefore concluded that these coefficients should be determined for each specific subject. However, the use of a simple two-point slope was found to be highly subject to error due to the effects of measurement error as discussed previously in section 3.3. Thirdly, in some cases, particularly those involving elderly subjects with non-compliant arteries or those under the influence of a vasodilators or any of a variety of drugs which affect arterial compliance, no reliable PWTT - SBP relationship was found to exist. In these cases, it was concluded that it is not possible to make estimates of changes in LOP based on monitoring PWTT. As a result it was concluded that measured PWTT appears to be a useful technique for adaptive control of a surgical tourniquet system when utilized for a large portion of subjects, while producing completely erroneous results in others. It was therefore concluded that, if the Chapter 7: Improvements for Application as a Clinical Tool <| 37 technique is to be considered for routine clinical use, it must be improved to function at a clinically acceptable level of reliability by employing PWTT only on appropriate subjects and during appropriate periods during surgical procedures. In view of these results and conclusions, the development of an improved clinically acceptable system was undertaken as described in section 7.2. The implementation and clinical evaluation of the improved system was beyond the scope of the thesis research and but the algorithm proposed for the system was applied to both retrospective and prospective clinical data for evaluation as described in sections 7.2.3 and 7.2.4. 7.2 Integration of an Oscillometric BP Monitor with the PWTT Measurement System 7.2.1 Description of the integrated system One possible improvement of the prototype system which was explored to achieve a clinically acceptable level of performance for the adaptation of surgical tourniquet systems was the integration of the PWTT measurement system with a BP monitor such as typically found in most operating rooms. A new algorithm was developed for this integrated system and the algorithm functions as follows. Periodic SBP values obtained from the BP monitor are utilized to "build" and continually update a model relating SBP to measured PWTT and to check its accuracy in predicting SBP changes. The validity and accuracy of the model is then be evaluated before allowing adaptive control of the surgical tourniquet system to be undertaken based on measured PWTT. Subjects for which a "good" PWTT-SBP relationship exists are identified and the tourniquet operated adaptively based on measured changes in PWTT. Subjects for which a "poor" relationship exists are also identified such that the surgical tourniquet system remains at a fixed pressure. Details regarding the algorithm of this proposed integrated system are given in section 7.2.2. Chapter 7: Improvements for Application as a Clinical Tool -]38 Slope and intercept values are determined taking into account each successive data point corresponding to a measured SBP value from the periodic BP monitor and fitting a least-squared error linear regression model to the accumulated PWTT-SBP pairs. The resulting slope and intercept values determined from the linear regression model are then based on a number of sample points and constantly updated with each new SBP-PWTT pair, rather than two discrete points as described previously. As a result, the effect of measurement errors is averaged out over time. Once large enough BP changes have been observed and sufficient samples are taken to adequately model the arterial path for a given subject and pulse sensor location, the model is tested for accuracy in predicting BP values as determined by the periodic BP monitor, as described in section 2.4.2. If the model proves accurate, it can then be utilized from that point in time onward to estimate LOP, allowing adaptive control the tourniquet cuff pressure. The slope of the linear regression model is utilized as the variable C in equation 3.6, in conjunction with a preoperatively measured LOP/SBP ratio, thereby allowing the continuous estimation of LOP and minimum effective tourniquet pressure through the ongoing measurement of PWTT. By closely monitoring the accuracy of the adaptive linear regression model in predicting each new SBP reading from the periodic BP monitor, as described in section 7.2.2, physiological changes or BP estimates which bring into question the accuracy of the model and the stability of the PWTT/BP relationship in the subject are detected and cause the surgical tourniquet system to revert to a fixed pressure mode. The potential advantages of such an integrated system over a system based only on measured PWTT are great. The system only instigates adaptive control of the tourniquet in the case that a good linear regression model relating SBP to PWTT is established based on a number of sample points. If a poor relationship is found to exist then the cuff remains set at a preset fixed pressure assuring a blood free surgical field. In addition, the system works to Chapter 7: Improvements for Application as a Clinical Tool -j gg develop a set of linear regression model parameters based on each individual arterial path with intercepts and slope coefficients that are specific to the location of the pulse sensor. This allows the sensor to be located at any point on the subject without requiring direct knowledge by the system as to where that point is, as long as it remains constant. If by chance the sensor is moved during the procedure, the subsequent (PWTT.SBP) pairs obtained for each periodic BP reading from the BP monitor will be found to be inconsistent with the linear regression model and the system will be returned to fixed pressure operation. 7.2.2 Algorithm for adaptive tourniquet operation In developing an algorithm for the operation of an adaptive surgical tourniquet system based on the integrated system proposed in section 7.2.1, the following assumptions were made: • readings from the oscillometric monitor are available every four minutes and sufficient error detection is included in the system to reject erroneous BP readings that are excessively high or low; and • an LOP reading is taken at the outset of each procedure and paired with measured SBP to allow determination of an LOP/SBP ratio. The flow chart for operation of the proposed improved system is presented in Fig 7.1. Based on these assumptions, the integrated system described above is configured such that a minimum 10 mmHg shift in SBP must be observed as recorded by the oscillometric monitor in order to allow development of a linear regression model relating PWTT to SBP. Following this minimum shift, a linear regression model is calculated which generates the least sum of squared errors after each valid (PWTT.SBP) pair based on all valid pairs to that point. For each revised linear regression model, the coefficient of correlation is determined. Each new oscillometric BP reading is compared with the corresponding estimate from the linear Chapter 7: Improvements for Application as a Clinical Tool regression model to the previous reading and used to compute the "error" associated with the model. START SET TOURNIQUET AT FIXED PRESSURE RECORD (SBP. PWTT) FOR EACH PERIOOIC BP MEASUREMENT Yes x CALCULATE LINEAR REGRESSION MODEL BASED ON ALL SAMPLE POINTS WAIT FOR NEXT PERIODIC BP MEASUREMENT Yes s UPDATE MODEL TO INCLUDE NEW POINT SET TOURNIQUET AT FIXED PRESSURE UNTIL TWO CONSECUTIVE SAMPLES MEET THS REQUIREMENT Yes S INSTIGATE ADAPTIVE CONTROL OF TOURNIQUET SYSTEM BASED ON LOP ESTIMATED FROM MEASURED PWTT USING CURRENT LINEAR REGRESSION MODEL Fig. 7.1: Flow chart for operation of the improved system. Chapter 7: Improvements for Application as a Clinical Tool The system then operates in one of two modes, fixed or adaptive. 141 FIXED MODE - In fixed mode the integrated system maintains the tourniquet cuff at a fixed pressure selected by the surgical staff at the outset of the procedure. This is set arbitrarily at 300 mmHg, the most common fixed pressure used in lower limb surgeries by the surgeons participating in this study. ADAPTIVE MODE - The integrated system enters this mode when a linear regression model relating PWTT and SBP has been established consisting of at least three points over a minimum SBP shift of 10 mmHg which has predicted the last sample to an error of less than 10 mmHg and has a correlation coefficient of 0.8 or greater. This correlation coefficient threshold has been determined by qualitative analysis of the sample data obtained from clinical testing. In this mode the required tourniquet pressure is estimated by eq. 3.6, PTOURNIQUET = LOPINITIAL + POFFSET + (LOPINITIAL/ SBPINITIAL ) * A P W T T * C where a safety offset POFFSET is utilized as an additional offset to account for minor variables and very sudden increases in BP which might otherwise cause leakage under the cuff, C is the slope of the current linear regression model and A P W T T is the difference between the PWTT value at the preoperative LOP measurement and the current value. A value of 25 mmHg was selected for POFFSET- During the periods between the periodic BP monitor readings, the cuff reacts to adapt to large increases in SBP as predicted using the established PWTT model. No adaptation based on decreased SBP estimates is made until confirmed by the next BP monitor reading. Once in adaptive mode the system reverts to fixed mode in the Chapter 7: Improvements for Application as a Clinical Tool -|42 case that a sample estimate is found to be in error by greater than 10 mmHg or if the coefficient of correlation of the linear regression model falls below 0.8. When back in fixed mode, the integrated system will return to adaptive operation when two consecutive periods occur where the error and correlation once again fall within the above limits. In the case that a BP monitor reading is unavailable or found to be erroneous, the safety offset is increased to 50 mmHg for the next sample period. If this occurs for two consecutive periods, the system reverts to fixed mode. In addition a minimum pressure value is set beyond which the pressure is not decreased. A value of 190 mmHg was selected based on current clinical practice. For periods during which PWTT values are unavailable due to motion artifact and for other reasons, the system is configured such that the cuff is slowly increased in pressure until either PWTT values once again become consistently available or the predetermined fixed pressure level is reached. 7.2.3 Results of retrospective application of apparatus to data from clinical trials The algorithm detailed in 7.2.2 was applied to the data obtained from initial clinical testing detailed in Chapter 6. The analysis of a sample case from this study is presented in Fig. 7.2. A complete table of results is included in Appendix E. Chapter 7: Improvements for Application as a Clinical Tool SEE NOTES 1-9 CASE B D 7 [11 [21 131 [41 151 161 171 [81 PWTT SBP SAMPLE SHIFT CORR ERROR MODE Pcuff 320 108 1 0 X x fixed 300 323 108 2 1 X x fixed 300 317 107 3 8 X x fixed 300 290 115 4 8 X x fixed 300 300 115 5 15 X x fixed 300 279 122 6 15 X x fixed 300 276 120 7 21 0.97 6 PWTT 220 271 128 8 21 0.96 -6 PWTT 230 266 128 9 21 0.97 -4 PWTT 230 268 127 10 21 0.97 -1 PWTT 230 269 127 1 1 21 0.97 -1 PWTT 230 274 127 12 21 0.97 -3 PWTT 230 262 135 13 28 0.96 -7 PWTT 250 263/231 NOTES: [11 PWTT in msec as measured using experimental system [21 Systolic blood pressure in mmHg as measured by monitor [31 Sample number in order from start of procedure,samples at 4 minute intervals [41 Maximum shift in SBP since beginning of measurement (no model until at least 10) 15] Coefficient of correlation of LR model formed using all data to current sample point, x if no model [61 Error in using last LR model to estimate current BP as determined using O.R. monitor, x if no model [71 Mode of control based on guidelines listed in text [8] Cuff pressure in mmHg, set at 300 for fixed mode or (SBP) • 1.6 + 25 in PWTT mode. All values rounded up to next 10 mmHg [91 Time averaged cuff pressure in mmHg. Total average /adaptive mode average. Fig. 7.2: Sample case for retrospective testing of the operation of the improved system. The results for cases which provided usable results are summarized in Table 7.1. Chapter 7: Improvements for Application as a Clinical Tool 144 TOTAL % TIME IN AVERAGE AVERAGE CASE TOTAL TIME ADAPTIVE ADAPTIVE PRESSURE PRESSURE (minutes) MODE TIME (minutes) MODE (mmHg) DURING ADAPTIVE MODE (mmHg) BD1 12 4 33% 273 220 BD2 36 12 33% 267 200 BD3 16 0 0% 300 300 BD4 28 20 71% 224 194 BD5 28 8 29% 274 210 BD7 52 28 54% 263 231 BD8 24 8 25% 270 195 BD9 64 16 25% 287 248 BD10 32 4 13% 286 190 BD11 12 0 0% 300 300 Table 7.1: Summary of results from retrospective study. The results of applying the algorithm in Fig. 7.1 are as follows. The system could operate in adaptive mode for 25 of the 77 periods recorded for short arthroscopic procedures (BD1-BD12), or 33 percent of the total elapsed tourniquet inflation time. The average tourniquet pressure during these periods was 216 mmHg, based on a LOP/SBP ratio of 1.6 with an overall time-averaged cuff pressure of 273 mmHg resulting in a 27 mmHg decrease in cuff pressure compared to the fixed pressure which otherwise would have been used. These values could be expected to vary in practice depending on the actual LOP/SBP ratio and the safety offset utilized, but, they are sufficient to suggest that a significant reduction in cuff pressure could be achieved in practice by integration of oscillometric BP and PWTT monitoring. Chapter 7: Improvements for Application as a Clinical Tool 145 When applied to cases RM2 and RM3 (those for which very poor correlation existed between PWTT and SBP) the theoretical system remained in fixed mode for the duration of the procedure. 7.2.4 Results of application of apparatus to data from prospective cases In order to further test the proposed apparatus, additional clinical testing was carried out on 8 patients as listed in table 7.2. SUBJECT ID A G E SEX ANESTHETIC PROCEDURE BD2.1 34 F GENERAL BONE STAPLE REMOVAL BD2.2 45 M GENERAL BONE STAPLE REMOVAL BD2.3 37 M GENERAL BONE STAPLE REMOVAL BD2.4 23 M GENERAL ARTHROSCOPIC KNEE SURGERY BD2.5 73 F GENERAL ARTHROSCOPIC KNEE SURGERY BD2.6 39 M GENERAL ACL RECONSTRUCTION BD2.7 51 F GENERAL ARTHROSCOPIC KNEE SURGERY BD2.8 18 M GENERAL ACL RECONSTRUCTION Fig. 7.2: Sample subjects for prospective study of proposed apparatus. This additional testing was undertaken to include the recording of the time intervals between BP measurements by the periodic BP monitor and PWTT measurement updates in order to accurately simulate the operation of the apparatus as envisioned. The same experimental protocol was used as described previously. The results are summarized in Table 7.3. Chapter 7: Improvements for Application as a Clinical Tool 146 TOTAL % TIME IN TIME AVERAGE CASE TOTAL TIME ADAPTIVE ADAPTIVE AVERAGED PRESSURE (minutes) MODE TIME MODE PRESSURE DURING (minutes) (mmHg) ADAPTIVE MODE (mmHg) BD2.1 38 0 0% 300 300 BD2.2 52 30 57% 247 208 BD2.3 39 25 64% 260 238 BD2.4 35 15 43% 259 224 BD2.5 40 25 63% 239 203 BD2.6 25 3 12% 300 300 BD2.7 44 26 59% 255 224 BD2.8 79 39 49% 267 233 TOTAL 352 163 46% 264 224 Table 7.3: Summary of results from prospective study. A representative sample of the data and analysis for the prospective study is included in Appendix E. It was found in these cases that the apparatus was able to operate adaptively for 46 percent of the overall time for the cases (163 minutes out of 352 minutes) with a time-averaged tourniquet pressure of 264 mmHg, representing a 12 percent reduction from a fixed pressure of 300 mmHg. Two additional laboratory trials were conducted to ascertain the validity of using the apparatus on: the toe of a subject; and the same arm as the BP cuff. This was desired in order to simulate upper limb surgery in which the arterial pulse sensor must either be located on the same limb as the BP cuff or alternatively located elsewhere on the body, most likely the toe. These two laboratory trials were designed to conform to the procedures and conditions of the operating room with the exception of the application of anesthetics. The Chapter 7: Improvements for Application as a Clinical Tool -| 47 results of these two trials showed that the system could operate effectively with the pulse sensor applied in these locations. 7.3 Considerations Regarding the possible Use of the Pulse Signal from a Pulse Oximeter for Pulse Arrival Detection A pulse oximeter is a device which monitors S 0 2 (blood oxygen saturation) by analyzing the ac component of the light transmitted through blood carrying tissues by two pulsed LEDs of different wavelengths [37,54,58,60]. Many pulse oximeters also provide a photoplethysmographic analog output which can be utilized to observe the real-time pulse waveform at the application site of the pulse oximeter probe. Pulse oximeters are standard equipment in the operating room and the sensing probe is routinely attached to the finger, or less frequently the earlobe, toe or a fold of skin on the upper arm. Using the photoplethysmographic analog output of a pulse oximeter to detect the arrival of the arterial pulse, it may prove possible to develop a PWTT measurement system based solely on the output of the operating room ECG monitor and the operating room pulse oximeter. This would allow the measurement of PWTT without using any additional sensors by simply connecting to the analog outputs from existing operating room monitoring equipment. The signal from some pulse oximeters may be found to be unsuited for preoperative measurement of LOP since the pulse oximeter will tend to alarm and otherwise react unpredictably as the pulse becomes very weak and therefore a dedicated probe may be required for this purpose. The use of the signal from the pulse oximeters for the measurement of PWTT is of significant advantage compared to the use of a dedicated pulse sensor. The following sections outline the possible benefits and limitations of a pulse oximeter-based system. Chapter 7: Improvements for Application as a Clinical Tool <| 43 Signal Reliability Since the pulse oximeter is standard equipment in the modern operating room, operating personnel will not begin a procedure until a proper output is obtained from the unit. As such they will re-apply the probe or otherwise adjust the sensor position to ensure a good signal is obtained and everything is working properly before proceeding. As pulse acquisition can sometimes be difficult, this would greatly assist in the acceptance of an adaptive tourniquet system, since the operators would not be required to apply or adjust another probe from a "new" system to ensure proper operation. Once the normal operating room equipment is applied and functioning properly such that the operation can begin, the PWTT measurement system will be assured of an acceptable pulse signal. Safety The use of a pulse oximeter output would remove the need for any electrical contact between the PWTT measurement system and patient. As such, safety issues involving leakage currents, microshock hazards, and heating of the sensor will be significantly reduced. Cost Since there is no need for a dedicated arterial pulse sensor or the necessary controlling hardware and software, an oximeter-based system would be significantly less expensive than the use of a dedicated puJse sensor. Chapter 7: Improvements for Application as a Clinical fool -|4g Timing As it is necessary to accurately time events based on the pulse signals, any large delay in the photoplethysmographic output of the pulse oximeter from real time due to signal processing could introduce a large error in PWTT measurement. If such a delay is constant and relatively small, it would not adversely affect estimation of blood pressure changes. However, if the delay is variable, it could be perceived to be a changing transit time and therefore a change in blood pressure. Therefore, the signal processing circuitry for every pulse oximeter for which the system is approved must be thoroughly evaluated to ensure that lengthy or varying delays in the waveform from real-time are not possible. Compatibility Although in the past analog outputs have been commonplace on most operating room equipment, this is changing, particularly as there is a move to integrate all operating room monitoring equipment into large integrated monitoring stations. Where discrete units are utilized, output signals are often digital and of varying format, and thus a list of acceptable pulse oximeters may have to be specified for an oximeter-based system. Significant potential exists for the integration of the signals from a pulse oximeter in order to facilitate continuous LOP estimation for adaptive control of a surgical tourniquet system. By using equipment that is present in most operating rooms and is used by the anesthetist in most surgical procedures the apparatus necessary to make continuous LOP estimations by measuring PWTT would be inexpensive to implement. No additional sensors and little new equipment would be needed. All PWTT measurements and LOP estimations made by the Chapter 7: Improvements for Application as a Clinical Tool ^ gg tourniquet controller could be made from signals output from the existing operating room equipment. Chapter 8: Conclusions and Recommendations CHAPTER 8 CONCLUSIONS AND RECOMMENDATIONS 151 8.1 Conclusions Many injuries which result from the use of surgical tourniquet systems are known to be related to cuff inflation pressure. Previous work has been done in the development of adaptive tourniquet pressure controllers which regulate the cuff pressure near the minimum required to occlude the limb in order to minimize the average pressure applied by the tourniquet to the underlying tissues. However, practical problems have plagued early adaptive tourniquet systems and precluded their routine clinical use. In particular, no clinically acceptable means has been found to accurately and non-invasively estimate a subject's BP continuously or at sufficiently frequent intervals. The purpose of this research was to investigate the use of continuously measured PWTT in order to continuously estimate LOP for the purpose of the adaptive control of a surgical tourniquet system. This investigation culminated in the development of an apparatus integrating the signals from existing operating room equipment which could be inexpensively implemented for continuously estimating the minimum tourniquet pressure to serve as the basis for an adaptive surgical tourniquet system suitable for routine clinical use. In order to perform laboratory and clinical testing to analyze the relationship between changes in PWTT and changes in LOP, a prototype PWTT measurement system was developed. This was the first goal in this thesis research and the following summarizes the results of this work: Chapter 8: Conclusions and Recommendations 152 (1) a non-invasive photoplethysmographic sensor which was designed, built and tested was found to be effective for producing an arterial pulse signal indicative of the arterial pulse sensed at a peripheral point on the body; (2) a filter and noise rejection algorithm was developed and tested for processing the arterial pulse signal prior to analysis and proved to be effective in both attenuating small amplitude noise and rejecting signals corrupted by significant motion artifact; (3) a detection algorithm based on analysis of the second derivative of the arterial pulse signal waveform was developed and tested to identify the arrival of the leading edge of the arterial pressure pulse; and (4) a system was designed, built and tested to provide continuous beat-to-beat PWTT measurements to an estimated accuracy of +/-7 msec, for typical transit times of 250-350 msec, utilizing the arterial pulse sensor which was developed and the output from an ECG monitor. Study of the arterial pulse and associated noise signals from the developed photoplethysmographic sensor led to the conclusion that a sophisticated digital signal processing approach was not warranted in solving the noise problem. A simple analog filter was proposed and implemented for suppressing noise outside the pulse signal band and a series of noise rejection routines were implemented in the system software to identify periods in which the arterial pulse signal was corrupted by noise. Given the nature of the measurements required from the signal, it was determined that intermittent noisy periods could be ignored and PWTT measurements could resume when a suitably clean signal became available once again. Chapter 8: Conclusions and Recommendations -] 53 It was determined that shape matching and cross correlation techniques were unsuited for determining the arrival of the arterial pulse wave due to the diversity of pulse shapes that could be expected from different subjects and sensor locations and due to the presence of waves reflected from the artery-arteriole junctions. A simple detection scheme was implemented based on identification of the peak acceleration point which characterizes the upturn of the arterial pressure waveform at the arrival of the pulse wave. This was found to work well in practice. More complex detection schemes might prove even more effective [69,70], but any increases in accuracy would likely be small and largely nullified by the running average of eight samples which was utilized to generate a smoother measurement system output. The second goal of this thesis research involved the development of an algorithm based on measured PWTT for possible use in the control of an adaptive surgical tourniquet system and this goal was achieved as follows: (1) an algorithm was developed which relates measured changes in PWTT to estimated changes in LOP; (2) based on the results of laboratory and clinical testing it was determined that the measurement of PWTT alone is insufficient to ensure accurate estimation of changes in LOP because of inter-patient variations in arterial physiology; (3) an improved apparatus was designed for estimating LOP and minimum safe tourniquet pressure for each patient based on the integration of continuously measured PWTT values with periodic BP estimates from a typical operating room BP monitor; Chapter 8: Conclusions and Recommendations 154 (4) the results of retrospective application of the algorithms for the improved apparatus to the data obtained from 16 representative subjects indicated that it was able to adaptively control surgical tourniquet pressure under clinically realistic circumstances resulting in the use of tourniquet pressures that were on average 27 mmHg or 9 percent lower than tourniquet pressures typically set at an arbitrary level of 300 mmHg and that it thus may provide an inexpensive and significant safety improvement for surgical tourniquet systems; and (5) the results of a prospective study on 8 additional subjects indicated that the integrated system operated in adaptive mode for 46 percent of the total time of surgical tourniquet application with an estimated time averaged tourniquet pressure decrease of 12 percent compared to a fixed pressure of 300 mmHg. Based on these results it was concluded that an apparatus that continuously estimates LOP for each patient by integrating measured PWTT with periodic estimates of the patient's BP from a readily available BP monitor may prove useful in the adaptive control of a surgical tourniquet system. By developing a linear regression model relating measured PWTT to periodic BP measurements, an adaptive surgical tourniquet system can develop specific slope coefficients for each subject and pulse sensor location. Analysis of the correlation between BP estimates made using the model coefficients and estimates generated periodically by the BP monitor will facilitate the identification of subjects for which a poor PWTT/LOP relationship exists; for these subjects the tourniquet cuff can be inflated to a fixed pressure. For subjects for which the correlation exceeds preset thresholds, the surgical tourniquet system can be operated adaptively. Utilizing the algorithms developed in this project, it was estimated that by using PWTT to estimate LOP in this manner surgical tourniquet system could operate in Chapter 8: Conclusions and Recommendations -| 55 adaptive mode for over 30-46 percent of the total time of tourniquet application with an estimated overall time averaged decrease in tourniquet pressure of 9-12 percent over a fixed pressure of 300 mmHg for lower limb surgeries and an average pressure decrease of 25-28 percent during the periods of adaptive operation. These estimates are based on study of 24 subjects who were undergoing lower limb surgeries where a minimum fixed pressure of 300 mmHg was used. Of these cases, 8 represent a proactive application of the algorithms in the clinical setting. The apparatus which was developed allows the use of any arterial path for PWTT measurement without prior knowledge by the estimation routines as to the arterial pulse sensor location. While a photoplethysmographic sensor applied under the cuff generally produces signals from which the pulse arrival can be detected, problems with inverted waveforms and difficulty with application may limit use in the clinical setting. Therefore, the use of a digital pulse sensor is most suitable and use of the photoplethysmographic signal obtained from the analog output of a pulse oximeter may then prove an excellent alternative to the use of a dedicated pulse sensor, provided problems involving compatibility with the signal outputs from various different pulse oximeter machines can be resolved. The photoplethysmographic signal from some pulse oximeters may also be suitable for performing the initial LOP measurement by placing the probe on a digit of the operative limb and looking for the disappearance of the pulsatile waveform as the tourniquet cuff is inflated. However, some pulse oximeters may generate erratic signals as they search for the weakened pulse and most units will generate audible "weak signal" alarms which will need to be manually silenced. Chapter 8: Conclusions and Recommendations -| 55 In spite of these minor drawbacks with the use of the photoplethysmographic signal from a pulse oximeter in place of a dedicated pulse probe, significant potential exists for the integration of the signals from an ECG monitor, BP monitor and pulse oximeter in order to facilitate continuous LOP estimation for adaptive control of a surgical tourniquet system. By using equipment that is present in most operating rooms and is used by the anesthetist in most surgical procedures the apparatus necessary to make continuous LOP estimations by measuring PWTT would be inexpensive to implement. No additional sensors and little new equipment would be needed. All PWTT measurements and LOP estimations made by the tourniquet controller could be made from signals output from the existing operating room equipment. 8.2 Contributions of this Research The primary contributions of this research relate to the development of a surgical tourniquet system which can adapt tourniquet cuff inflation pressure in respect to changes in the patient's LOP resulting in the effective use of lower tourniquet pressure and thus potentially increasing patient safety. These contributions are summarized as follows: (1) the development and validation of a prototype PWTT measurement system which can be implemented in a simple and inexpensive manner through the use of patient monitoring equipment commonly used in the operating room; (2) the identification of problems preventing the use of PWTT based algorithms to accurately estimate BP in elderly patients or those undergoing drug therapy and in the use of non patient specific model coefficients or coefficients obtained through the determination of a two-point slope based on pre- and post-exercise BP and PWTT values; and Chapter 8: Conclusions and Recommendations 157 (3) the development and clinical validation of an apparatus for the continuous estimation of LOP and thereby minimum safe tourniquet inflation pressure based on the integration of periodic BP estimates available from a periodic BP monitor commonly utilized during surgery with continuously measured PWTT values. 8.3 Recommendations and Topics for Further Research Prior to development of an adaptive tourniquet system based on measured PWTT it is necessary to obtain representative data from a larger sample of subjects in order to facilitate selection of safety offset values for tourniquet inflation pressure and select the decision thresholds based on developed linear regression models for switching between fixed and adaptive mode. In this research, the scope of the project and the time constraints placed upon it limited extensive clinical testing. The algorithms developed for the improved system in this project, while adequate for the purpose of investigating the clinical viability of such a system are largely subjective, based on the small number of samples in this study and may not be optimized for use in general practice. In particular, the selection of the safety offset value utilized for operation in adaptive mode must be investigated further. Clinical study in this area will require cooperation on the part of the surgeon in allowing the tourniquet to be deflated slowly at some point during the procedure to determine at what tourniquet inflation pressure leakage into the surgical field occurs. It may also be possible to study the selection of safety offsets in the laboratory by monitoring the volume change over time of an exsanguinated limb on which a tourniquet has been applied and inflated to a pressure equal to the sum of LOP and a selected safety offset value. Any subsequent increase in the volume of the limb distal to the tourniquet may indicate the ooze of blood under the cuff. Chapter 8: Conclusions and Recommendations 158 It is also necessary to study the effects of varying tourniquet pressure over the duration of a surgical procedure in order to determine if the use of an adaptive tourniquet at a lower, continually changing pressure results in a lower probability of injury than the use of a higher fixed inflation pressure. Fundamental questions remain concerning the effects of tourniquet pressure, inflation time and surface pressure distribution on underlying neural damage. No quantitative information exists relating tourniquet parameters to neural damage in humans as only a qualitative understanding has been developed based mostly on animal studies and sporadic surgical case reports. Further investigation is also recommended to ascertain the effect of various drugs, primarily vasoconstrictors and vasodilators on PWV as measured in the peripheral arteries and vascular system. Such study must be undertaken in conjunction with a person knowledgeable in the field of medicine and anesthetic practice. It might prove possible based on data from an expanded study to build a data base relating a variety of physiological parameters such as pulse rate, rate of change of pulse rate, pulse amplitude, rate of change of pulse amplitude, pulse shape, PWTT and rate of change of PWTT such that a pattern recognition approach could be applied in conjunction with linear regression modeling for improved estimation of LOP over a greater portion of surgical procedures and for a greater number of subjects. Investigation of these aspects appears to be warranted as the results of the research described in this thesis suggest that safer, cost effective surgical tourniquet systems might be developed based on the measurement of PWTT. 159 REFERENCES [1] Frank Hinman, "The Rational Use of Tourniquets," International Abstract of Surgery, vol.81, no.5, pp.357-366,1945. [2] Hubert S. Reid, et al.," Tourniquet Hemostasis," Clinical Orthopedics and Related Research, pp. 230-234, August, 1983.McLaren, C.H. [3] Rorabeck, "The Pressure Distribution Under Tourniquets," The Journal of Bone and Joint Surgery, vol. 67-A, no. 3, pp. 433-438, 1985. [4] Andrew Muirhead, Raymond J. Newman, "A Low Pressure Tourniquet for the Lower Limb," Injury, vol. 17, no.8, pp. 17-18, 1986. [5] Gordon McConnel, "The Use of Impedance Plethysmography to Predict the Onset of Blood Flow Beneath a Tourniquet Cuff," M.A.Sc. Thesis, University of British Columbia, Vancouver, Canada, April 1988. ;6] R. Sanders, "The Tourniquet: Instrument or Weapon?," H a n d , vol.5, no.2, pp.119-123, 1973. 7] L. Klenerman, "Tourniquet Time -How Long ?," Hand, vol.12, no.3, pp.231-234, 1980. 8] L. Klenerman et al.," Systemic and Local Effects of the Application of a Tourniquet," Journal of Bone and Joint Surgery, vol.62-B, no.3, pp.385-388, 1980. 9] Raymond J. Newman, Andrew Muirhead, "A Safe and Effective Low Pressure Tourniquet," The Journal of Bone and Joint Surgery, vol. 68-B, no.4, pp.625-628, 1986. 10] O.Levy, Y. David , et al., "Minimal Tourniquet Pressure to Maintain Arterial Closure in Upper Limb Surgery," The Journal of Hand Surgery, vol.18B, no.2, pp.204-206,1993. 11] James A. McEwen, Robert W. Mcgraw," An Adaptive Tourniquet for Improved Safety in Surgery," IEEE Transactions on Biomedical Engineering, vol.29, no.2, pp. 122-128, 1982. 12] James A. McEwen, G.F. Auchinleck, "Advances in Surgical Tourniquets," AORN Journal, vol. 36, no.5, pp. 889-896, 1982. 13] James A. Shaw, David G. Murray, "The Relationship Between Tourniquet Pressure and Underlying Soft Tissue Pressure in the Thigh," The Journal of Bone and Joint Surgery, vol. 64-A, no. 8, pp. 1148-1152, 1982. 14] Stephen C. Grice, et al., "Intravenous Regional Anesthesia: Evaluation and Prevention of Leakage Under the Tourniquet," Anesthesiology, vol. 65, no. 3, pp. 316-320, 1986. 15] M. J. Breault et al., "Internal Pressure Distribution in Limbs Encircled By Pneumatic Surgical Tourniquets," CMBEC-15-CCCGB Toronto, pp. 47-49, 1989. 160 [16] Brent Graham et al., "Occlusion of Arterial Flow in the Extremities at Subsystolic Pressures Through the Use of Wide Tourniquet Cuffs," Clinical Orthopedics and Related Research, no. 286, pp.257-261, 1993. [17] Ho Jung Kang et al.," Blood Gas and Electrolyte Changes After Tourniquet Application in Total Knee Replacement Surgery," Yonsei Medical Journal, vol. 32, no.2, pp. 153-158, 1992. [18] Albert G. Crenshaw et al.," Wide Tourniquet Cuffs More Effective at Lower Inflation Pressures," Acta Orthop Scand, vol. 59(4), pp.447-451, 1988. [19] M.R. Moore, S.R. Garfin, A.R. Hargens," Wide Tourniquets Eliminate Blood Flow at Low Inflation Pressures," The Journal of Hand Surgery, vol.12A, no.6, pp.1006-1011,1987. [20] Robert A. Pedowitz et al.," The Use of Lower Tourniquet Inflation Pressures in Extremity Surgery Facilitated by Curved and Wide Tourniquets and an Integrated Cuff Inflation System," presented in part at the 34th annual meeting of the Orthopedic Research Society, 1988. [21] Mark E. Miller, "Estimation of Limb Occlusion pressure By Adaptation of Oscillometry for Surgical Tourniquet Control," M.A.Sc. Thesis, University of British Columbia, Vancouver, Canada, December 1989. [22] L. Klenerman, "Tourniquet Paralysis," Journal of Bone and Joint Surgery (Br), vol 65, pp. 374-375, 1983. [23] Shaw, D.G. Murray, "The Relationship BetweenTq Pressure and Underlying Soft Tissue Pressure in the Thigh," Journal of Bone and Joint Surgery (Am), vol 64, pp. 1148-1152, 1982. [24] McArdle, F.I. Katch, V.L. Katch, "Exercise Physiology, Second Edition," Lea & Febiger Publishing, Philadelphia, 1986. [25] Van De Graaff, S.I. Fox, "Concepts of Human Anatomy and Physiology, Third Edition," W.M.C. Brown Publishers, Dubuque, IA, 1992. [26] L.B.Rowell et al., "Disparities Between Aortic and peripheral Pulse Pressures Induced by Upright Exercise and Vasomotor Changes in Man," Circulation, vol.XXXVII, pp.954-964, 1968. [27] Edwin J. Kroeker, Earl H. Wood," Comparison of Simultaneously Recorded Central And Peripheral Arterial pressure Pulses During Rest, Exercise and Tilted Position in Man," Circulation Research, vol.lll, pp.623-632, 1955. [28] O'Rourke, "Function of Conduit Arteries," Arterial Vasodilation - Mechanisms and Therapy, Lea & Febiger Publishing, Philadelphia, 1993. [29] O'Rourke, "Wave Travel and Reflection in the Arterial System," Arterial Vasodilation -Mechanisms and Therapy, Lea & Febiger Publishing, Philadelphia, 1993. 161 [30] Nichols et al. ."effects of Age and of Hypertension on Wave Travel and Reflections," Arterial Vasodilation - Mechanisms and Therapy, Lea & Febiger Publishing, Philadelphia, 1993. [32] Malcolm Carruthers, Peter Taggart, "Validation of a New, Inexpensive, Non-invasive Miniaturized Blood-Pressure Monitor," Journal of Ambulatory Monitoring, vol.1, no.2, pp.163-170, 1988. [33] Toshihiko Ishimitsu, Shigeru Yagi, "Report on Clinical Findings on the Usefulness of Blood Pressure Monitor of Watch Type model BP-100 (a Product of CASIO)," produced for the Department of Circulatory Internal Medicine, Dokkyo University, Japan, May 1991. [34] Ichio Matsumoto, "Report on Clinical Findings on the Usefulness of Blood Pressure Monitor of Watch Type model BP-100 (a Product of CASIO)," produced for Saitama Medical School, Japan, May 1991. [35] O'Rourke, R. Kelly, A. Avolio, "The Arterial Pulse," Lea & Febiger Publishing, Philadelphia, 1992. [36] Stoelting, "Pharmacology and Physiology in Anesthetic Practice," J.B. Lippincott Company, Philadelphia, 1987. [37] J.G. Webster (Editor), "Medical Instrumentation Application and Design - Second Edition," Princeton N.J., Houghton Mifflin Company, 1992. [38] J.W. Hole, "Human Anatomy and Physiology - Third Edition," Dubuque Iowa, Wm. C. Brown Publishers, 1984. [39] J.C. Bramwell, A.V. Hill, "The Velocity of the Pulse Wave in Man," The Proceedings of the Royal Society, no. 93, pp. 298-306, 1922. [40] Gershon Weltman, George Sullivan, Dale Bredon," The Continuous Measurement of Arterial pulse Wave Velocity," Medical Electronics and Biological Engineering, vol.2, pp. 145-154, 1964. [41] B.Gribbin, A. Steptoe, P Sleight," Pulse Wave Velocity as a Measure of Blood Pressure Change," in Psychophysiology, vol.13, no.1, pp.86-90, 1976. [42] Hemingway," The Extensability of Human Arteries," The quarterly Journal of Medicine, vol.21, pp. 489-498, 1928. [43] "Continuous Photometric Blood Pressure Theory and Technology", paper produced by Sentinel Monitoring, Indianapolis IN, in relation to their products ARTRAC™ 7000 and ARTRAC™ 5000. [44] L.A. Geddes et al.," Pulse Transit Time as an Indicator of Arterial Blood Pressure," Psychophysiology, vol.8, no.1, pp.71-74, 1981. 162 [45] J. Marymont et al., "Evaluation of the ARTRAC Noninvasive, Non-occlusive and Continuous Blood Pressure Monitor in Cardiac Patients," study by the Division of Anesthesiology, Northwestern University, Evanston Hospital, Evanston IL. [46] F.E. Yates, Laurel A. Benton," Blood Pressure Levels and Variance Assessed by Ambulatory Monitoring: Optimal Parameters," Annals of Biomedical Engineering, vol.18, pp.377-405, 1990. [47] Nrtish V. Thakor," Reliable R-Wave Detection from Ambulatory Subjects," Biomed. Sci. Instrum., 14:67-72, 1978. [48] Jiapu Pan, Willis J. Tompkins," A Real-Time QRS Detection Algorithm," IEEE Transactions on Biomedical Engineering, vol.32, no.3, pp.230-236, 1985. [49] M.L. Ahlstrom, W. J. Tompkins," Digital Filters for Real-Time ECG Signal Processing Using Microprocessors," IEEE Transactions on Biomedical Engineering, vol.32, no.9, pp.708-713, 1985. [50] Quizhen Xue, Yu Hen Hu, W.J. Tompkins," Neural-Network-Based Adaptive Matched Filtering for QRS Detection," IEEE Transactions on Biomedical Engineering, vol.39, no.4, pp.317-329, 1992. [51] Tompkins, Webster, "Design of Microprocessor Based Instrumentation," Englewood Cliffs, Prentice-Hall Co., 1981. [52] Thakor, J.G. Webster, W.J. Tompkins, "Optimal QRS Detector," Med. Biol. Eng. Comput. vol21, pp. 343-350, 1983. [53] Hokaanson et al.," An Electrically Caalibrated Plethysmograph for Direct Measurement of Limb Blood Flow," IEEE Transactions on Biomedical Engineering, BME-22, pp.25-29, 1975. [54] Yitzhak Mendelson, Burt D. Ochs," Noninvasive Pulse Oximetry Utilizing Skin Reflectance Photoplethysmography," IEEE Transactions on Biomedical Engineering, vol.35, no.10, pp.798-805, 1988. [55] K.G. Murphy, J.A. Secunda, M.A. Rockoff," Severe Burns from a Pulse Oximeter," Anesthesiology, vol 73, pp.350-352, 1990. [56] Tod B. Sloan," Finger Injury by an Oxygen Saturation Monitor Probe," Anesthesiology, vol 68, no.6, pp.936-938, 1988. [57] Weijia Cui, Lee E. Ostrander, Bok Y. Lee," In Vivio Reflectance of Blood and Tissue as a Function of Light Wavelength," IEEE Transactions on Biomedical Engineering, vol.37, no.6, pp.632-639, 1990. [58] Y. Mendelson et al.," Design and Evaluation of a New Reflectance Pulse oximeter Sensor," Medical instrumentation, vol.22, no.4, pp.167-173, 1988. 163 [59] Electromedical Equipment, C22.2, no.601.1 - M90, Ottawa: Canadian Standards Association, 1990. [60] Y. Mendelson, Brenda L. Yocum," Noninvasive Measurement of Arterial Oxyhemoglobin Saturation with a Heated and Non-heated Skin Reflectance Pulse Oximeter Sensor," Biomedical Instrumentation and Technology, pp.472-479, November/December 1991. [61] "Silicon Photodiodes," Product catalog for Centronic inc., Newbury Park, CA. [62] Mortin Benthin et al.," Calculation of pulse wave Velocity Using Cross Correlation -Effects of Reflexes in the Arterial Tree," Ultrasound in Medicine and Biology, vol.17, no.5, pp.461-469, 1991. [63] R.L. Scheaffer, J.T. McClave, " Probability and Statistics for Engineers - Second Edition," Boston MA., Duxbury Press, 1986. [64] Electromedical Equipment, C22.2 no.125-M1984, Ottawa: Canadian Standards Association, 1984. [65] Douglas L. Wood et al. ," Cold Pressor test as a Predictor of Hypertension," Hypertension, vol.6, no.3, pp. 301-306, 1984. [66] W.R. Harlan, R.K. Osborne, A. Graybiel," Prognostic Value of the Cold Pressor Test and the Basal Blood Pressure," American Journal of Cardiology, pp.683-687, May 1964. [67] M.H. Ellestad, "Stress Testing - Principles and Pactice, Edition 3," F.A. Davis Company, Philadelphia,1986. [68] Yaginuma, "Modification of Wave Travel and Reflection by Vasodilator Therapy," Arterial Vasodilation - Mechanisms and Therapy, Lea & Febiger Publishing, Philadelphia, 1993. [69] Chen-Chin Lee, Roger G. Mark," Analysis of arterial Waves by the Single Pulse Response Method," IEEE Transactions on Biomedical Engineering, vol.40, no.8, pp.833-836, 1993. [70] J.Y. Lee, J.C.Lin, " A Microprocessor Based Noninvasive Arterial Pulse Wave Analyzer," IEEE Transactions on Biomedical Engineering, vol.32 no.6, pp.451-455, 1985. [71] W.M. Nichols, M.F. O'Rourke, "McDonald's Blood Flow in Arteries - Third Edition," Lea & Febiger Publishing, Philadelphia, pp. 77-86, 1990. 164 APPENDIX A - CIRCUIT DIAGRAMS FOR EXPERIMENTAL SYSTEM 165 LED CURRENT DRIVER CIRCUIT f j AMBIENT LIGHT R E J E C T I O N C O N T R O L HSO.2 f j S A M P L E A N D HOLD C O N T R O L HSO.1 TO AMPLIFICATION A N D A N A L O G FILTER S T A G E S S E N S O R PHOTODIODE PHOTODIODE BIASING AMBIENT LIGHT R E J E C T O N S A M P L E Af^ H O L D " S T A G E _ PHOTODIODE BIASING CIRCUIT 166 4-POLE LOW PASS FILTER 3-POLE HIGH PASS FILTER OUTPUT FROM . ARTERIAL PULSE / SIGNAL SAMPLE AND HOLD STAGE ARTERIAL PULSE WAVEFORM FILTERING AND AMPLIFICATION CIRCUIT •5V ECG SIGNAL , INPUT ' ECG ANALOG , OUTPUT GROUND HSIO R-WAVE DETECTOR CIRCUIT PULSE WIDTH MODULATED PORT OUTPUT t-5V O h 5.6K -AA/V 10uF . PWM OUTPUT TO VOLTAGE CONVERSION CIRCUIT 78M08UC V O L T A G E REGULATOR +8V +5V GND -8V SYSTEM POWER SUPPLY AND VOLTAGE REGULATION CIRCUIT 168 APPENDIX B - SOFTWARE LISTINGS AND FLOWCHARTS 169 FLOW CHART FOR SYSTEM SOFTWARE DEFINE VARIABLES AND INITIALIZE STACK POINTERS DELAY LNTIL LED PULSE CYCLE HAS FINISHED INCREMENT PULSE COUNTER AND STACK POINTERS CALCULATE TIMING POINTS FOR NEXT LED PULSE CYCLE QRS_1 SET UP HGH SPEED INPUT PORT TO CAPTURE AND FLAG POSITIVE GONG TRANSmCN FROM R-WAVE DETECTOR INPUT CIRCUIT STORE RESULTS OF LATEST A/D SAMPLE CONVERSION IN STACK SLOPEJ DETERMINE FIRST DERIVATIVE OF THE PULSE WAVEFORM AS THE DIFFERENCE BETWEEN THE CURRENT SAMPLE MAGMTUDE ^ AND THE THIRD PREVIOUS SAMPLE MAGMTUDE CALL SLOPE 2 SLOPE_2 DETERMINE SECOND DERIVATIVE OF THE PULSE WAVEFORM AS DIFFERENCE BETWEEN THE CURRENT SLOPE VALUE AND THE SECOND PREVIOUS SLOPE VALUE AN LOOP CONTINUE ON PAGE 2 PAGE 1 OF 3 170 WW LOOP CONTINUED FROM PACE I THRESHOLDSET CALCULATE THRESHOLD VALUES FOR NOISE REJECTION IN THE NEXT PULSE PERIOD BASED ON THE PULSE CHARACTERISTICS OF THE LAST PERIO0 7 \ VALI DITY_C H EC K , CHECK THE PLLSE \ CHARACTERISTICS OF THE I PREVIOUS PERIOD AGAINST THE / THRESHOLDS SET FROM THE / SECOND PREVIOUS PERIO0 / IS LAST SAMPLE . VALID ? RETURN YES > NO T RETURN SET INVALID SAMPLE FLAG CALCULATE AVG TIME STORE LATEST VAUD TIME IN STACK AND CALCULATE AVERAGE PWTT FROM LAST 8 VAUO TIMES. CALCULATE THRESHOLD VALUES FOR SECOND AVERAGE / SECON D_AVERAG E CALCULATE AVERAGE FROM PWTT VALUES THAT FALL WtTHN THRESHOLDS SET IN MAIN AVERAGING ROUTINE PAGE 2 OF 3 CONTINUE ON PAGE3 171 CONTINUED FROU PAGE 2 NO % CLEAR RUMWG MAXIMUM DCWArtARD ACCELERATION VAFBABLE Jfc. CALL REWMAX ] NEW_MAX_3 KEEPS RLWNG TRACK OF UMMLM VALUE OF SAMPLE WAVEFORM SINCE LAST R-WAVE. RESET AFTER EACH R-WAVE. WHEN NEW MINIMUM IS DETECTED. UPDATES UMNO REGISTER WTH TIME OF MAXIMUM ACCELERATION ON ASSOCIATED SECTION OF WAVEFORM AS OETERMNED BY NEW MAX 2 / HAS R-WAVE V D X . / BEEN DETECTE  S . SNCE LAST QRS_2 DETERMNE PWTT FROM LAST CYCLE. CLEAR COUNTERS * D RLWJNG MAGNTUOEiSLOPE/ ACCELERATION VARIABLES CALL CHARACTERISTICS | CHARACTERISTICS \ DETERMNE PULSE \ CHARACTERISTICS FROM LAST l CYCLE BASED ON ARRIVAL TIME DETERMNED BY LAST TIMING / REGISTER LPOATE FROM / NEW MAX 3 / \ / VWT LNTIL TIME TO GO TO LED PLLSE ROUTIME CALL SPULSE SPULSE LOAD TIMNG S C O N C E FOR SH LED. AMBIENT LIGHT REJECTION AND AO CONTROL UNES FOR NEXT PLLSE PERI CO PAGE 3 OF 3 S O E T L I S T . T X T $ D E B U G $ I N C L U D E ( . . \ I N C L U D E \ 8 0 C 1 9 6 K D . I N C ) ; O U T P U T I S TO PWM » A N A L O G OUT Y I E L D I N G . 0 1 9 5 V = ; C O N F I G U R E D FOR EPROM PROGRAMMING ; F I N A L V E R S I O N U T I L I Z E D FOR C L I N I C A L O . R . T E S T I N G 3 m s e c ; l a s t u p d a t e d may 1 6 , 1 9 9 4 . *************************** ; S T A R T / d e f i n e t e m p o r a r y v a r i a b l e s * ; v a l u e s a r e a r b i t r a r y R S E G A T O O l C h S T A R T O : P E A K F L A G : L E D S T A R T : i n e A A : A B : A C : A D : A E : A F : A G : V L E D M A X : DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ***************************************** ; s t a r t o f L E D t i m i n g p e r i o d ; f l a g f o r p e a k d e t e c t r o u t i n e , a n d i n i t . ; e n d o f L E D t i m i n g p e r i o d , c a l l p u l s e r o u t ; H S O t i m i n g p o i n t s . / s t o r a g e l o c a t i o n f o r L E D v o l t a g e s i g n a l / •max imum a l l o w a b l e v a l u e S _ H E A D L O : DSW 1 S HEAD M I N U S : DSW 1 ; h e a d p o i n t e r f o r c i r c u l a r s t a c k , l o b y t e ; t a i l p o i n t e r f o r c i r c u l a r s t a c k S S _ H E A D : DSW 1 S S HEAD M I N U S : DSW 1 ; h e a d p o i n t e r f o r 1 s t d e r i v a t i v e s t a c k ; t a i l p o i n t e r f o r 1 s t d e r i v a t i v e s t a c k A V E R A G E 1 _ M I N U S : DSW 1 A V E R A G E 1 P L U S : DSW 1 / • t h r e s h o l d s f o r t w o s t e p a v e r a g e S C O U N T E R : DSW ; p u l s e p e r i o d c o u n t e r V _ M I N : T _ M I N : V : V A R R : DSW DSW DSW DSW / •m in imum v a l u e o f p u l s e w a v e f o r m ; t i m e o f r u n n i n g m i n o f p u l s e w a v e f o r m , - r u n n i n g m a g n i t u d e a t l a s t z e r o s l o p e , - r u n n i n g m a g n i t u d e a t a r r i v a l S H E A D H I : DSW ; h e a d p o i n t e r f o r c i r c u l a r s t a c k , h i b y t e P a g e 1 173 S O F T L I S T . T X T T E M P I : DSW 1 ; t e m p o r a r y w o r d v a r i o u s u s e s n n ™ ^ R o : D S W 1 c o u n t e r s f o r a r r i v a l r o u t i n e s P C 0 U N T E R 2 : DSw 1 R S E G A T 5 1 52 5 3 54 5 5 5 6 57 58 5 9 5 1 0 5 1 1 5 1 2 5 1 3 5 1 4 5 1 5 5 1 6 0 0 5 0 H DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ; 1 6 e l e m e n t c i r c u l a r s t a c k " S " ; p u l s e v o l t a g e v a l u e s 5 5 1 5 5 2 5 5 3 5 5 4 5 5 5 5 5 6 5 5 7 5 5 8 5 5 9 5 5 1 0 5 5 1 1 S S I 2 5 5 1 3 5 5 1 4 5 5 1 5 5 5 1 6 DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 ; 1 6 e l e m e n t c i r c u l a r s t a c k " S S " ; l s t d e r i v a t i v e o f p u l s e w a v e f o r m T I M E 1 DSW . 1 ; 9 0 h T I M E 2 DSW 1 T I M E 3 DSW 1 ; 8 e l e m e n t T I M E 4 DSW 1 T I M E 5 DSW 1 ; 9 8 T I M E 6 DSW 1 ; 9 a T I M E 7 DSW 1 ; 9 c T I M E 8 DSW 1 ; 9 e P a g e 2 S O F T L I S T . T X T W O RKREG: D S L P M A G : DSW 1 P R I S E : DSW . 1 M A G Z E R O : DSW 1 PMAG C O P Y : DSW 1 T H M A G 7 5 : DSW 1 T H M A G 1 2 5 : DSW 1 ; w o r k r e g i s t e r f o r t h r e s h o l d c a l c s ; a O , a l , a 2 , a 3 ; b - p m a g n i t u d e o f l a s t p u l s e a4 ; b - p r i s e t i m e o f l a s t p u l s e a 6 ; r i s e b e l o w z e r o ( d c o f f s e t p o i n t ) , - m a g n i t u d e c o p y f o r c h a r a c t e r i s t i c s r o u t n ; 7 5 % m a g n i t u d e t h r e s h o l d ; 1 2 5 % m a g n i t u d e t h r e s h o l d H S I _ M 0 D E _ C 0 P Y : I O C 0 _ C O P Y : I 0 C 1 _ C 0 P Y : T H R I S E 7 5 : Q R S _ T I M E _ N E W : I B I : PWTT: T H P E A K 1 2 5 : T H P E A K 7 5 : T H M A G Z E R 0 2 5 : T H R I S E 1 2 5 : N E W 1 : 0 L D 1 : W O R K I N G : S _ H E A D _ O L D : S S _ H E A D _ 0 L D : T I M E R _ H E A D : T H M A G Z E R 0 7 5 : S _ H E A D _ 0 L D 2 : S S _ H E A D _ 0 L D 2 : S _ H E A D _ 0 L D 3 : T E M P A C E L : A V E R A G E 1 : O L D _ A C C E L : A C C E L : V M A X : X X : TEMP R O L L : T M I N : T O T P E A K S : T F L A G : W 0 R K I N G 2 : D S B DSB D S B 1 1 1 DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSW DSB DSB DSW DSB DSB DSW ; c o p y v a r i a b l e s f o r QRS p o r t s e t u p 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 75% r i s e t i m e t h r e s h o l d t i m e r l v a l u e a t c u r r e n t QRS i n t e r b e a t i n t e r v a l a s c a l c u l a t e d t r a n s i t t i m e a s c a l c u l a t e d 125% t i m e t o p e a k t h r e s h o l d 75% t i m e t o p e a k t h r e s h o l d 25% d e f l e c t i o n t h r e s h o l d 125% r i s e t i m e t h r e s h o l d w o r k i n g v a r i a b l e f o r s l o p e l , 2 , - p o i n t e r t o l a s t e n t e r e d s , - p o i n t e r t o l a s t e n t e r e d s s , - p o i n t e r t o l a s t e n t e r e d ; 7 5 % d e f l e c t i o n t h r e s h o l d ; f i r s t a v e r a g e o f t i m e s ; o l d a c c e l e r a t i o n / • c u r r e n t a c c e l e r a t i o n / •maximum o f i n v e r t e d p u l s e w v f r m ; c o u n t o f s a m p l e s w i t h i n a v g t h r / • t e m p o r a r y h o l d e r f o r ; r o l l o v e r c o u n t a s p u l s e r o u t i n e / • s t a r t s u s e d i n NEW MAX ; t i m e o f p u l s e m i n i m u m ;# o f p e a k s d u r i n g s a m p l e p e r i o d ; s l o p e , a l s o r e - u s e d t o s t o r e S P a g e 3 i?5 S O F T L I S T . T X T S S _ M A X _ V A L : TEMP3: T E M P 2 : W I N D O W _ O P E N : W I N D O W _ C L O S E : T A R R : A V E R A G E 2 : V A L I D F L A G : C O R R E C T _ F L A G : L A S T _ A V E R A G E : M I S S E S : P S L O P E : e V M I N : . * * * * * * * * * * * * * * * * * * * * * * * * * * * * ^ . * * * * * * * * * * * * * * * * * * * * • * • * s e t u p i n t e r u p t v e c t o r s * * * * * * * * * * * * * * * * * * * * * * * * * / i n t e r r u p t a u t o m a t i c a l l y v e c t o r t o dOOOh l o c a t i o n s - s e t u p ; j u m p s t o a p p r o p r i a t e r o u t i n e s a t t h e s e a d d r e s s e s DSW 1 / r u n n i n g m a x i m u m s l o p e D S B 1 ; t e m p h o l d f o r p u l s e DSW 1 ; t e m p h o l d f o r p u l s e DSw 1 / w i n d o w i n g p o i n t s DSw 1 DSw 1 ; t i m e o f p u l s e a r r i v a l DSW 1 / c o r r e c t e d a v e r a g e D S B 1 , - c l e a r e d i f s a m p l e i n v a l i d DSB 1 DSW 1 , - a v e r a g e f o r t w o s t e p p r o c e s s D S B 1 ; c o u n t o f c o n s e c u t i v e b a d a v e r a g e s DSB 1 / • c h a r a c t e r i s t i c s l o p e o f l a s t p u i s DSW 1 ; v a l u e o f m i n i m u m p o i n t o n w a v e f r m C S E G A T 0 D 0 0 0 H BR N O _ R O U T I N E NOP NOP : t i m e r o v e r f l o w h a n d l i n g C S E G A T 0 D 0 2 0 H BR L E D _ C U R R E N T _ C H E C K NOP NOP r a / d d o n e h a n d l i n g ; s t a r t m a i n p r o g r a m l i s t i n g C S E G A T 0 2 0 1 8 h DCB O E B h ; S E T C H I P C O N F I G U R A T I O N B I T C S E G A T 0 2 0 8 0 h ; A U T O S T A R T A D D R E S S ON R E S E T D I L D S P , # 0 0 3 0 0 H ; i n i t i a l i z e s t a c k p o i n t e r L D S _ H E A D L O , I 0 0 5 0 H ; i n i t i a l i z e c i r c u l a r s t a c k L D S S _ H E A D , # 0 0 7 0 H / - p o i n t e r s P a g e 4 S O F T L I S T . T X T L D T I M E R H E A D , # 0 0 9 0 H C L R B T F L A G A N D B I O C 0 _ C O P Y , # 1 1 1 1 1 1 1 0 b S T B IOCO C O P Y , I O C 0 H S I _ T I M E R _ R E G I S T E R _ C L E A R : g s L D P E A K F L A G , H S I _ T I M E J B S I O S l , 7 , H S I _ T I M E R R E G I S T E R C L E A R ; d i s s a b l e QRS d e t e c t ; r t n t o c l e a r H S I t i m e r r e S C A L L QRS 1 ; s e t up H S I l i n e s ; n o t a c t i v a t e d y e t ; * * * * * * * * * * * * * * * * * * * * m a i n p r o g r a m l o o p * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ; * * * * * * * * * * * * * * * * * * + + + + + ^ + + + + + + + + + + i + + + + + + + i ^ ^ ^ + i + i t + + + ^ + + + + + + ^ + i + + j t + + ^ ; r u n s i n b a c k g r o u n d a s t i m e d b y L E D p u l s i n g R M A I N : : * * * * * L D S T A R T O , # 0 4 0 0 d L D L E D S T A R T , # 0 4 4 d ; l o a d s t a r t t i m e w i t h 4 0 0 d * * * * * * * * w a s t e t i m e u n t i l new S T A R T O i s r e a c h e d ( w h i l e HSO i s e x e c u t e d ; r r o m r e g i s t e r s s e t u p i n p r e v i o u s p e r i o d ) * * * * * * * * * * * * * * * * * * * * * * r * * * * * * * * * W A I T 2 : W A I T 3 : CMP T I M E R 1 , S T A R T O ; * * * * * w a i t f o r e n d o f B L E ; * * * * * d e l a y p e r i o d WAIT 3 S J M P W A I T 2 CMP T I M E R 1 , S T A R T O ; i f r o l l o v e r BGT W A I T 2 CMP T I M E R 1 , S T A R T O ; n o r m a l c a s e B L E WAIT 3 ;*****************+*********** * * * * * * * * * * * * * * * * * + + i. + J r i . + + ^ + + + l t J r + + * * * * * * * ORB I O C 0 _ C O P Y , # 0 0 0 0 0 0 0 1 b S T B I O C 0 _ C O P Y , I O C O I N C P C O U N T E R 1 ; e n a b l e H S I . O i n p u t ; i n c r e m e n t p u l s e c o u n t e r P a g e 5 1 7 ? S 0 F T L I S T . T X T L E D S T A R T , S T A R T O , # 3 3 8 4 d / d e t e r m i n e s t a r t t i m e f o r new p u l s e / r o u t i n e * * * * * s e t u p p u l s e r o u t i n e s w i t c h i n g t i m e s * * * * * * * * * * * * * * * * * * * * * * * * ADD A A , L E D S T A R T , # 5 0 d / a m b i e n t o f f ADD A B , L E D S T A R T , # 6 0 d / L E D o n ADD A C , L E D S T A R T , # 1 0 0 d ; s / h o n , r e a d V i e d ADD A D , L E D S T A R T , # 1 4 0 d / s / h o f f ADD A E , L E D S T A R T , # 1 6 0 d / L E D o f f ADD A F , L E D S T A R T , # 2 0 0 d / r e a d p u l s e v o l t a g ADD A G , L E D S T A R T , # 2 2 2 d / t u r n o n a m b i e n t ADD S T A R T O , S T A R T O , # 3 7 5 0 d / s e t new b e g i n t i m e / i n STARTO * * i n c r e m e n t c i r c u l a r s t a c k p o i n t e r s * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * L D S HEAD O L D 3 , S HEAD 0 L D 2 L D S HEAD 0 L D 2 , S HEAD OLD L D S S _ _ H E A D _ 0 L D 2 , S S _ H E A D _ O L D L D S HEAD O L D , S H E A D L O / p o i n t e r t o l a s t e n t e r e d ADD S H E A D L O , S H E A D L O , # 2 / i n c r e m e n t p o i n t e r b y t w o CMP S H E A D L O , # 0 0 7 0 H / c h e c k i f h i t t o p o f s t a c k J N E OK 1 / i f n o t , s k i p n e x t l i n e L D S _ H E A D L O , # 0 0 5 0 H / i f s o , s e t b a c k a t s t a r t ADD S H E A D H I , S H E A D L O , # 0 0 0 1 H / s e t h i g h b y t e a d d r e s s ADD SS H E A D , S H E A D L O , # 0 0 2 0 H / s e t new SS HEAD ADD S S _ H E A D _ O L D , S _ H E A D _ O L D , # 0 0 2 0 h / s e t new S S HEAD OLD s t h e • n e w e s t p u l s e v a l u e a c q u i r e d * * * * * / s t o r e n e w e s t p u l s e S T B AD R E S U L T L O , [ S H E A D L O ] / v a l u e a t h e a d o f S T B A D _ R E S U L T _ H I , [ S _ H E A D H I ] / S s t a c k L D T E M P I , [ S H E A D L O ] / c o p y t o t e m p o r a r y SHR T E M P I , # 6 / s h i f t r i g h t 6 s u c h P a g e 6 \n S O E T L I S T . T X T ST T E M P I , [ S H E A D L O ] ; t h a t 10 b i t v a l u e i s @ R H S ; c o p y b a c k t o s t a c k ; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * / l o c a t i o n o f m a i n c a l c u l a t i o n r o u t i n e c a l l s ; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * L C A L L S L 0 P E _ 1 L C A L L S L 0 P E _ 2 ; p o s t QRS r o u t i n e s CMPB C O R R E C T _ F L A G , # 1 BNE N O T _ R I G H T _ A F T E R S C A L L T H R E S H O L D _ S E T NOT R I G H T A F T E R : i f p e r i o d i m m e d i a t e l y f o l l o w i n g Q R S , c a l l <<<<<< CMPB C 0 R R E C T _ F L A G , # 2 BNE N O T _ F I R S T _ A F T E R _ Q R S S C A L L V A L I D I T Y _ C H E C K C L R B T O T P E A K S C L R B P E A K F L A G N O T _ F I R S T _ A F T E R _ Q R S : CMPB C O R R E C T _ F L A G , # 3 BNE N O T _ S E C _ A F T E R _ Q R S S C A L L C A L C U L A T E A V G T I M E ; i f p e r i o d i m m e d i a t e l y / • f o l l o w i n g Q R S , c a l l <<<<<< a n d c l e a r p e a k c o u n t a n d a s s o c . f l a g i f 2 p e r i o d i m m e d i a t e l y f o l l o w i n g Q R S , c a l l <<<<<< N O T _ S E C _ A F T E R _ Q R S : CMPB C O R R E C T _ F L A G , # 4 B N E N O T _ T H I R D _ A F T E R _ Q R S S C A L L S E C O N D _ A V E R A G E N O T J T H I R D A F T E R Q R S : * * * * * * * * * * * * * * * w i n d o w e d c a l l s * * * * * * * * * * * * * * * * ; i f 3 p e r i o d i m m e d i a t e l y ; f o l l o w i n g Q R S , c a l l ;<<<<<< / o t h e r w i s e g o o n ! ! CMP P C O U N T E R 1 , # 1 5 D B L T NOT I N WINDOW i s p u l s e c o u n t < o p e n I N WINDOW: ; o t h e r w i s e P a g e 7 S O F T L I S T . T X T L C A L L NEW_MAX L C A L L N E W _ M A X _ 3 L C A L L E X T R A P E A K S N O T _ I N _ W I N D O W : * * * * * * * * * * * * * * * * * * J B S ; i f s k i p p e d Q R S _ C A L L : NO QRS C A L L : S J M P S C A L L Q R S _ 2 NOP I O S 1 , 7 , Q R S _ C A L L ; j u m p t o Q R S _ 2 i f H S I f l a g b i t s e t ; t o c a l c u l a t e t i m e s N O _ Q R S _ C A L L ; o t h e r w i s e s k i p ; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ; * * * * * * * * * * * * * * * * * * * * * + * * * * * * * * * * * * * * * * * * * * + + + + + + + + i + + ^ + i t j t + + ] l r J t + + + J r + i + j t + + + + . * + * * * w a s t e t i m e u n t i l t i m e t o g o t o p u l s e r o u t i n e a g a i n * * * * * * * * * * * * CMP T I M E R 1 , L E D S T A R T W A I T 4 : W A I T 5 : W A I T 6 : * * * * * * * * * * * CMP B L T W A I T 5 BE W A I T 6 S J M P WAIT4 CMP T I M E R 1 , L E D S T A R T BGT WAIT4 T I M E R 1 , L E D S T A R T B L T W A I T 5 NOP S J M P S P U L S E / w a s t e t i m e u n t i l t i m e t o / g o t o l e d p u l s e r o u t i n e / n o r m a l c a s e / c a l l p u l s e r o u t i n e t o s e t / u p HSO CAM r e g i s t e r s S J M P R M A I N / r e t u r n t o t o p a n d w a i t / f o r c o m p l e t i o n o f HSO ; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * S U B R O U T I N E S ; ****** ******** * * * j ^ E R r j p T g E Ryj_cj NQ**************** ******* ******* N O _ R O U T I N E : / r e t u r n f r o m i r r e l a v e n t i n t e r u p t s P a g e 8 S O F T L I S T . T X T R E T NOP NOP L E D _ C U R R E N T _ C H E C K : ;check Vied from a/d converter CMPB A D _ R E S U L T H I , V L E D M A X : compare to max allowed S P U L S E S U B R 0 U T I N E * * * * * * * * * * * * * * * * * * * , * ^ ^ ^ /sets up pulse outputs and a/d timing through use of CAM r e g i s t e r and : d l f d , J ^ ' ^ ^ J ^ t 0 e v e n t f w i l l S o u r ;un er i r e c t i o n of t i m e r l S P U L S E : W A I T : E C K D I /disable interupts J B S I O S O ,6,W A I T ;make sure CAM opening L D B L D H S 0 COMMAND,#00000010b H S 0 _ T I M E , A A ;turn o f f ambient L D B L D HSO COMMAND,#00100000b H S 0 _ T I M E , A B ; led .on L D B L D HSO COMMAND,#00100001b H S 0 _ T I M E , A C ;s/h on L D B L D B L D AD_COMMAND,#00010111b HSO COMMAND,#00101111b H S O _ T I M E , A C ;set up a/d ;8 b i t conversion ;a/d Vied L D B H S 0_COMMAND,#00010001b ;s/h o f f /generate an interupt -;to s e r v i c e L E D C U R R E N T L D H S 0 _ T I M E , A D L D B L D •HSO COMMAND,#00000000b H S O _ T I M E , A E ; L E D o f f L D B AD_COMMAND,#00000000b ;set up a/d ;10 b i t conversion ;a/d l i n e 00 L D B HSO_COMMAND, #00101111b ;read pulse voltage Page 9 S O F T L I S T . T X T L D H S O _ T I M E , A F L D B H S O _ C O M M A N D , # 0 0 1 0 0 0 1 0 b ; t u r n o n a m b i e n t L D HSO T I M E , A G S J M P R M A I N / r e t u r n t o m a i n p r o g r a m G A I N A D J U S T 1 : RET NOP ; * * * * * * * * * * * * * * * * * * * * * N E W t ^ * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ^ ^ * ^ ^ ^ ^ NEW M A X : / l o o k s d o w n t u r n i n w a v e f o r m ( p o s i t i v e s l o p e ) . I f n o / d o w n t u r n t h e n c l e a r s r u n n i n g max a c c e l , o t h e r w i s e c a l l s / n e w max 2 W 0 R K I N G 2 , # 0 0 A L R E A D Y _ T U R N E D T E M P A C E L CMP BNE C L R R E T A L R E A D Y _ T U R N E D : L C A L L NEW_MAX_2 RET / h a s s l o p e t u r n e d d o w n ? / i f s o s k i p t o e n d NEW MAX 2 : / a n d b / L D B L D CMP B L T CMP B L E / L D B / o r b S T ST L D / d e t e r m i n e s p e a k a c c e l e r a t i o n a l o n g e a c h d o w n t u r n e d / p e i c e o f t h e s a m p l e w a v e f o r m . s a v e s i n p c o u n t e r 2 / a n d s t o r e s c o r r e s p o n d i n g m a g n i t u d e , i o p o r t l , # 1 1 1 1 1 1 0 1 b I O P O R T 1 , # 0 F F h W O R K R E G , [ S _ H E A D L O ] W O R K R E G , T H M A G Z E R 0 7 5 TOO F A R DOWN W O R K I N G , T E M P A C E L N O T _ M A X _ A C E L I O P O R T 1 , # 0 0 H i o p o r t l , # 0 0 0 0 0 0 1 0 b W O R K I N G , T E M P A C E L P C O U N T E R 1 , P C O U N T E R 2 V , [ S _ H E A D L O ] / l o a d m a g n i t u d e i n t o w o r k r e g / c o m p a r e t o 75% d o w n / i f f u r t h e r d o w n s k i p / o t h e r w i s e / c o m p a r e c u r r e n t a c c e l t o max / s k i p i f l o w e r / o t h e r w i s e / u p d a t e m a x , / 1 i m e , / a n d m a g n i t u d e P a g e 10 S O F T L I S T . T X T T O O _ F A R _ D O W N : N O T _ M A X _ A C E L : R E T • * * * * * * * * * * * * * * * * * * * * * NEW MAX 3 * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ; * * * * ~ ~ ; k e e p s r u n n i n g t r a c k o f m i n i m u m v a l u e o f s a m p l e w a v e f o r m , r e s e t b y Q R S 2 ; w h e n p e a k , u p d a t e s r e g s w i t h max a c c e l e r a t i o n p o i n t a l o n g d o w n t u r n e d ; s l o p e a s d e t e r m i n e d b y NEW MAX N E W _ M A X _ 3 : L D S U B CMP BGT ; a n d b S J M P NEW_LOW ; o r b L D L D ST ST L D W 0 R K I N G 2 , l O O O O O O l l l l l l l l l l B ; l o a d w o r k i n g w i t h f s d W 0 R K I N G 2 , [ S _ H E A D L O ] / s u b t r a c t c u r r e n t v a l u e t o g e t / i n v e r t e d w a v e f o r m W 0 R K I N G 2 , V M A X / c o m p a r e c u r r e n t mag t o l o w e s t NEW_LOW / i f n o t l o w e r , s k i p t o e n d i o p o r t l , i U l l l l l l O b NOT NEW LOW i o p o r t l , # 0 0 0 0 0 0 0 1 b T M I N , P C 0 U N T E R 1 V M A X , W 0 R K I N G 2 P C 0 U N T E R 2 , T A R R V , V A R R V M I N , [ S H E A D L O ] / o t h e r w i s e s t o r e c u r r e n t T / a n d i n v e r t e d m a g n i t u d e / a n d s t o r e t i m e o f l a s t s l o p e / d o w n t u r n . / s t o r e c u r r e n t v a l u e a s m i n i m u m N 0 T _ N E W _ L 0 W : R E T NOP ; * * * * * * * * * * * * * * * * * * * * * Q R S * * * * * * * * * * * * * * + * * * * * * * + + + * * * * * * * * * * * * * * * * * * * * * / s e t s u p H S I p o r t O t o c a p t u r e p o s i t i v e g o i n g t r a n s i t i o n f r o m QRS d e t e c t o r / c i r c u i t r y , d o e s n o t e n a b l e , t h i s i s d o n e i n m a i n p r o g r a m l o o p QRS 1: r l A N D B H S I _ M O D E _ C O P Y , # 1 1 1 1 1 1 0 0 b ORB H S I _ M O D E _ C O P Y , # 0 0 0 0 0 0 0 1 b S T B . H S I _ M 0 D E _ C 0 P Y , H S I _ M 0 D E ORB I 0 C 1 C O P Y , # 1 0 0 0 0 1 0 1 b t t / r i s i n g e d g e / i n t e r r u p t o n t i m e / r o l l o v e r o r H S I / s t a c k f i l l / e n a b l e pwm f o r pw P a g e 11 !o3 S O F T L I S T . T X T ; o u t p u t S T B I 0 C 1 _ C 0 P Y , I O C 1 R E T NOP NOP NOP j * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * Q R S _ 2 ;************************************************************************* * ; c a l l e d b y f l a g s e t o n d e t e c t i o n o f QRS p u l s e o n H S I O Q R S _ 2 : L D Q R S _ T I M E _ N E W , H S I _ T I M E / s t o r e c u r r e n t H S I . O t i m e a s NEW ; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * + + + + + + + + + + + + + + + + + + J f * * * * * * * * * S C A L L C H A R A C T E R I S T I C S / d e t e r m i n e f i d u c i a l p o i n t s f o r l a s t p e r . L D B C O R R E C T _ F L A G , # 1 / s e t f l a g f o r f i r s t a f t e r QRS t o s t a r t / p o s t QRS r o u t i n e c y c l e s ; * * * * * * * * * * * * * * * * * * * c j _ e a r r e g i s t e r s f o r n e x t c y c l e * * * * * * * * * * * * * * * * * * * * * * * / l o a d p u l s e m i n t r a c k e r w i t h a l l l ' s / r e s e t c o u n t e r / c l e a r r u n n i n g a c c e l e r a t i o n t r a c k e r L D V M A X , # 0 0 h C L R B T F L A G C L R P C 0 U N T E R 1 C L R T E M P A C E L c l r b i o p o r t l R E T NOP . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * . * * * * * * * * * * * * * * * * * * * * * * * * * * * s ] _ o p e i * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * — / d e t e r m i n e t h e f i r s t d e r i v a t i v e o f t h e p u l s e w a v e f o r m S L 0 P E _ 1 : C L R WORKING2 / s e t c u r r e n t WORKING v a l u e t o 0 0 L D 0 L D 1 , [ S _ H E A D _ O L D 3 ] / m o v e v a l u e s i n s t a c k t o L D N E W 1 , [ S _ H E A D L 0 ] / t e m p v a r i a b l e s P a g e 12 S O F T L I S T . T X T CMP N E W 1 , 0 L D 1 BGE P O S _ S L O P E S U B W O R K I N G 2 , O L D l , N E W l P O S _ S L O P E : S T W O R K I N G 2 , [ S S _ H E A D ] ADD W O R K I N G 2 , [ S S _ H E A D _ O L D ] n RET NOP ; * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * s l o p e 2*********************************** / d e t e r m i n e t h e 2 n d d e r i v a t i v e o f t h e p u l s e w a v e f o r m S L 0 P E _ 2 : C L R WORKING L D O L D 1 , [ S S _ H E A D _ O L D 2 ] L D N E W 1 , [ S S _ H E A D ] . CMP N E W 1 , 0 L D 1 B L E N E G _ S L O P E SUB A C C E L , N E W 1 , 0 L D 1 ADD W O R K I N G , A C C E L , O L D _ A C C E L ST A C C E L , O L D _ A C C E L N E G _ S L O P E : RET NOP . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * c a l c u l a t e a v g t i m e * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * / u p d a t e s t i m e r s t a c k p o i n t e r s a n d f i n d s a v e r a g e p w t t o v e r l a s t 8 s a m p l e s ; r u n i n 2 n d p o s t QRS i f v a l i d i t y c h e c k s p a s s e d / c o m p a r e new v a l u e t o l a s t / l e a v e s s z e r o i f p o s s l o p e / d e t e r m i n e n e g s l o p e / s t o r e i n s t a c k a s p o i n t e d / s u m l a s t t w o s l o p e s f o r c o m p a r i s o / s e t c u r r e n t WORKING v a l u e t o 00 / m o v e s t a c k v a l u e s t o t e m p / v a r i a b l e s / c o m p a r e new s l o p e t o l a s t / l e a v e s s s z e r o i f n e g a c c e l . / d e t e r m i n e p o s a c c e l / a d d t o l a s t f o r c o m p a r i s o n / s t o r e c u r r e n t a s l a s t C A L C U L A T E A V G T I M E : ADD T I M E R _ H E A D , # 0 2 d CMP T I M E R _ H E A D , # 0 A 0 h BNE N O T _ A T _ S T A C K _ T O P L D T I M E R _ H E A D , # 0 9 0 h NOT A T S T A C K T O P : / i n c r e m e n t s t a c k p o i n t e r , r e t u r n t o 90 / i f a t • t o p P a g e 1.3 S O F T L I S T . T X T S T T A R R , [ T I M E R _ H E A D ] ; s t o r e c u r r e n t p w t t a t h e a d C L R A V E R A G E 1 ; c l e a r a v e r a g e r e g i s t e r ADD A V E R A G E 1 , T I M E 1 ADD A V E R A G E 1 , T I M E 2 ADD A V E R A G E 1 , T I M E 3 ADD A V E R A G E 1 , T I M E 4 ADD A V E R A G E 1 , T I M E 5 ADD A V E R A G E 1 , T I M E 6 ADD A V E R A G E 1 , T I M E 7 ADD A V E R A G E 1 , T I M E 8 ; s u m l a s t 8 t i m e s SHR A V E R A G E 1 , # 3 ; d i v i d e b y 8 ADD A V E R A G E 1 _ P L U S , A V E R A G E 1 , # 5 ; s e t t h r e s h o l d s S U B A V E R A G E 1 _ M I N U S , A V E R A G E 1 , # 5 L D B C O R R E C T _ F L A G , # 4 ; s e t f l a g t o r u n c o r r e c t i o n r o u t i n e R E T NOP j * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * . * * * * * * * * * * * * * * * * * * * * * s e c o n d a v e r a g e * * * * * * * * * * * * * * * * * * * * * * * * ; r u n d u r i n g t h i r d c y c l e a f t e r Q R S , t h r o w s o u t h i g h / l o w v a l u e s a n d c a l c ; a v e r a g e f r o m t h e r e m a i n i n g v a l u e s , o n l y i f v a l i d i t y c h e c k p a s s e d S E C O N D _ A V E R A G E : a n d b i o p o r t l , # 1 1 1 1 1 0 1 1 b C L R A V E R A G E 1 C L R B X X ; c l e a r r e g i s t e r CMP B L E CMP B G E ADD I N C B NO 2 : T I M E 1 , A V E R A G E 1 _ M I N U S N 0 _ 2 T I M E 1 , A V E R A G E 1 _ P L U S N 0 _ 2 A V E R A G E 1 , T I M E 1 X X ; c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e ; i n c c o u n t e r a n d a d d i f i n s i d e CMP T I M E 2 , A V E R A G E 1 _ M I N U S B L E N 0 _ 3 CMP T I M E 2 , A V E R A G E 1 _ P L U S BGE N 0 _ 3 ADD A V E R A G E l , T I M E 2 ; c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e P a g e 14 S O F T L I S T . T X T I N C B NO 3 : X X ; i n c c o u n t e r a n d a d d i f i n s i d e CMP B L E CMP B G E ADD I N C B N O _ 4 : CMP B L E CMP BGE ADD I N C B N O _ 5 : CMP B L E CMP BGE ADD I N C B N O _ 6 : CMP B L E CMP BGE ADD I N C B N O _ 7 : CMP B L E CMP B G E ADD I N C B T I M E 3 , A V E R A G E 1 _ M I N U S NO_4 T I M E 3 , A V E R A G E 1 _ P L U S NO_4 A V E R A G E 1 , T I M E 3 X X T I M E 4 , A V E R A G E 1 _ M I N U S N O _ 5 T I M E 4 , A V E R A G E 1 _ P L U S N O _ 5 A V E R A G E 1 , T I M E 4 X X T I M E 5 , A V E R A G E 1 _ M I N U S N O _ 6 T I M E 5 , A V E R A G E 1 _ P L U S N O _ 6 A V E R A G E 1 , T I M E 5 X X T I M E 6 , A V E R A G E 1 _ M I N U S NO_7 T I M E 6 , A V E R A G E 1 _ P L U S NO_7 A V E R A G E 1 , T I M E 6 X X T I M E 7 , A V E R A G E 1 _ M I N U S NO_8 T I M E 7 , A V E R A G E 1 _ P L U S NO_8 A V E R A G E 1 , T I M E 7 X X ; c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e ^ i n c c o u n t e r a n d a d d i f i n s i d e : c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e ; i n c c o u n t e r a n d a d d i f i n s i d e ; c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e ; i n c c o u n t e r a n d a d d i f i n s i d e ; c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e ; i n c c o u n t e r a n d a d d i f i n s i d e ; c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e ; i n c c o u n t e r a n d a d d i f i n s i d e N0_£ CMP B L E CMP B G E T I M E 8 , A V E R A G E 1 _ M I N U S N 0 _ 9 T I M E 8 , A V E R A G E 1 _ P L U S NO 9 / c o m p a r e t o t h r e s h o l d , t o s s i f o u t s i d e P a g e 15 S O F T L I S T . T X T ADD A V E R A G E 1 , T I M E 8 I N C B X X ; i n c c o u n t e r a n d a d d i f i n s i d e N 0 _ 9 : L D B I O P O R T 1 , # 0 0 CMP B L T C L R B D I V U B ST L D B L D B X X , # 0 5 NOT ENOUGH S A M P L E S M I S S E S A V E R A G E 1 , X X A V E R A G E 1 , L A S T A V E R A G E I O P O R T l , x x ; c l e a r v a l i d s a m p l e d i s p l a y ; d i d > 3 / 8 v a l u e s g e t t o s s e d ? ; i f s o , n o o u t p u t , j u m p t o e n d ; l e d d i s p l a y w i l l s t a y c l e a r ; c l e a r m i s s c o u n t e r ; d i v i d e s u m b y n u m b e r o f v a l i d p o i n t s ; s t o r e a v e r a g e l a s l a s t o k o n e ; s h o w # o f v a l i d s a m p l e s P W M _ C O N T R O L , L A S T _ A V E R A G E ; o u t p u t c o r r e c t e d PWTT a v e r a g e NOT ENOUGH S A M P L E S : I N C B M I S S E S CMPB M I S S E S , # 1 0 D B L T NOPE N O P E : L D B R E T NOP C O R R E C T F L A G , # 5 ; m c m i s s c o u n t e r ; - m i s s e d 10 i n a r o w ? ; i f n o t , c o n t i n u e ; u p d a t e f l a g F o r p o s t QRS r o u t i n e s C H A R A C T E R I S T I C S ; c a l l e d f r o m Q R S 2 t o d e t e r m i n e t h e s l o p e , r i s e a n d m a g n i t u d e / c h a r a c t e r i s t i c s o f t h e l a s t p u l s e p e r i o d C H A R A C T E R I S T I C S : S U B P M A G , V A R R , V M I N / d e t e r m i n e b - p m a g n i t u d e o f p u l s e S U B P R I S E , T M I N , T A R R ; d e t e r m i n e a s s o c i a t e d r i s e t i m e L D M A G Z E R O , # 7 4 5 d ; l o a d DC o f f s e t l e v e l S U B M A G Z E R O , V M I N ; s u b t r a c t p u l s e m i n t o g e t d e f l e c t i o n SHR M A G Z E R O , # 1 ; d i v i d e d e f l e c t i o n b y 2 L D T H M A G Z E R 0 2 5 , # 7 4 5 d ; l o a d DC o f f s e t S U B T H M A G Z E R 0 2 5 , M A G Z E R O ; s u b t r a c t t o g e t t h r e s h o l d P a g e 16 S O F T L I S T . T X T ST P M A G , P M A G _ C O P Y S H L P M A G _ C 0 P Y , # 3 D I V U B P M A G _ C O P Y , P R I S E L D B P S L O P E , 0 0 a 4 h ; m a k e c o p y o f P M a g / m u l t i p l y b y 8 ; d i v i d e b y l o w b y t e o f r i s e t i m e P R I ; s t o r e l o w b y t e o f r e s u l t a s s l o p e RET NOP . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * T H R E S H O L D _ S E T . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ; s e t s new t h r e s h o l d s u s i n g c h a r a c t e r i s t i c s f r o m l a s t s a m p l e p e r i o d ; c a l l e d f r o m QRS 2 T H R E S H O L D S E T : SHR M A G Z E R O , # l M U L U W O R K R E G , M A G Z E R O , # 3 L D T H M A G Z E R 0 7 5 , # 7 4 5 d S U B T H M A G Z E R 0 7 5 , W O R K R E G M U L U WO R K R E G , P M A G , # 3 SHR W O R K R E G , # 3 L D THMAG7 5 , O O A O h M U L U W O R K R E G , P M A G , # 1 3 d SHR W O R K R E G , # 3 L D T H M A G 1 2 5 , O O A O h S U B T H R I S E 7 5 , P R I S E , # 6 d ADD T H R I S E 1 2 5 , P R I S E , # 6 d S U B T H P E A K 7 5 , T M I N , # 8 d C L R T H P E A K 1 2 5 ADD T H P E A K 1 2 5 , T M I N , # 8 d ; d i v a g a i n b y t w o t o g e t 1 /4 ; * 3 = 75% ; l o a d DC o f f s e t ; s u b t r a c t t o g e t 75% t h r e s h o l d / m u l t i p l y m a g n i t u d e b y 3 / d i v i d e b y 8 / s t o r e l o w w o r d a s 75% t h r e s h o l d / m u l t i p l y m a g n i t u d e b y 13 / d i v i d e b y 8 / s t o r e l o w w o r d a s 125% t h r e s h o l d / s u b t r a c t 1 8 m s e c f o r 75% r i s e / a d d 1 8 m s e c f o r 125% r i s e / s u b 24 m s e c f o r 75% t o p e a k t i m e / a d d 2 4 m s e c f o r 125% t o p e a k t i m e L D B C O R R E C T _ F L A G r # 0 2 h RET NOP . * * * * * * * * + * * * * * * * * * * * * * * * * * * * * * * * * * * * V A L I D I T Y _ C H E C K / c a l l e d f r o m a f t e r QRS t o c h e c k t h e c u r r e n t p u l s e c h a r a c t e r i s t i c s / a g a i n s t t h e t h r e s h o l d s s e t b a s e d o n t h e p r e v i o u s p u l s e p e r i o d P a g e 17 S O F T L I S T . T X T V A L I D I T Y _ C H E C K : l d b i o p o r t l , # 0 0 0 0 0 0 0 1 b CMP P M A G , T H M A G 7 5 ; c h e c k t h a t p u l s e mag i s w i t h i n l i m i t s B L E N O T _ V A L I D ; o t h e r w i s e b r a n c h t o e n d CMP P M A G , T H M A G 1 2 5 BGE NOT V A L I D l d b i o p o r t l , # 0 0 0 0 0 1 0 0 b CMP P R I S E , T H R I S E 7 5 B L E N O T _ V A L I D CMP P R I S E , T H R I S E 1 2 5 BGE N O T _ V A L I D l d b i o p o r t l , # 0 0 0 0 1 0 0 0 b CMP T M I N , T H P E A K 7 5 B L E N O T _ V A L I D CMP T M I N , T H P E A K 1 2 5 BGE NOT V A L I D ; c h e c k r i s e t i m e ; c h e c k t i m e o f p e a k o c c u r e n c e L D B I O P O R T 1 , # 0 0 0 0 0 0 1 0 b CMP T O T P E A K S , # 0 1 d BNE NOT V A L I D ; c h e c k f o r s p u r i o u s p e a k s ; i f we g e t t o h e r e , c u r r e n t s a m p l e i s g o o d cmpb b n e L D B RET n o t 2 n d : v a l i d f l a g , # 0 f h n o t _ 2 n d C O R R E C T F L A G , # 3 r i s f l a g s e t i n d i c a t i n g l a s t w a s g o o d ? r i f n o t s e t f l a g b u t d o n t u p d a t e r e g s ' i f a l l o k t h e n g o o n t o u p d a t e ' t i m e r e g i s t e r s a n d r e t u r n o r b i o p o r t l , # 0 0 0 0 0 1 0 0 b l d b v a l i d f l a g , # 0 f h l d b c o r r e c t _ f l a g , # 5 r e t ; s e t f l a g ; s k i p u p d a t e N O T _ V A L I D : ; o r b i o p o r t l , # 0 0 0 0 0 1 0 0 b o r b i o p o r t l , # 1 1 1 1 0 0 0 0 b L D B C O R R E C T _ F L A G , # 0 5 L D B V A L I D F L A G , # 0 0 r o t h e r w i s e c l e a r f l a g s o a v g . s k i p p e d r c l e a r v a l i d f l a g RET ; a n d r e t u r n P a g e 18 S O F T L I S T . T X T 1 9 0 NOP . * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * ; E X T R A _ P E A K S / r o u t i n e c a l l e d a f t e r e v e r y p u l s e a / d s a m p l e t o d e t e c t s p u r i o u s p e a k s ; i n t h e w a v e f r o m w h i c h m i g h t r e s u l t f r o m n o i s e o r m o t i o n a r t i f a c t E X T R A _ P E A K S : CMP T H M A G Z E R 0 2 5 , [ S _ H E A D L O ] B L T S E T _ F L A G CMP T H M A G Z E R 0 7 5 , [ S _ H E A D L 0 ] BGT P E A K _ C O U N T E R S J M P E N D P E A K P E A K _ C O U N T E R : J B C P E A K F L A G , 7 , E N D P E A K I N C B T O T P E A K S C L R B P E A K F L A G S J M P E N D P E A K S E T _ F L A G : L D B P E A K F L A G , # O F F h E N D P E A K : R E T NOP ; x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x / c o m p a r e c u r r e n t s m p l e t o 25% / ( a c t u a l t h r e s h o l d i s 50 %) / i f > , s e t f l a g / o t h e r w i s e c o m p a r e t o 75% / i f < , g o t o p e a k c o u n t e r / o t h e r w i s e r e t u r n / i f f l a g h a s b e e n c l e a r e d , e n d / o t h e r w i s e i n c c o u n t e r / a n d c l e a r f l a g / s e t f l a g i f >25% END P a g e 19 191 APPENDIX C - CHART TRACES FOR SYSTEM CALIBRATION Sample of chart trace for: sensor applied to forefinger; subject male, age 25; and trace scale = 25msec/div. Sample of chart trace for: sensor applied under forearm tourniquet cuff @ 200 mmHg; subject male, age 25; and trace scale = 25msec/div. Sample of chart trace for: sensor applied under front of thigh tourniquet cuff @ 200 mmHg; subject male, age 30; and trace scale = 25msec/div. Sample of chart trace for: sensor applied under front of thigh tourniquet cuff @ 250 mmHg; subject female, age 24; and trace scale = 25msec/div. '•WISE Sample of chart trace for: sensor applied to thumb; subject male, age 30; and trace scale = 25msec/div. — - —. i ! ! : 1 i > NOFTF) ; - j — : — ! — i — — i — 1 —.— ; M i l l ! 1 I M -i i i i 1 I i i ' ' 1 i - i—i i i 1 i i i i i i ! i i i ^ i - U J . !:! -L -!—i-Ml i a—L . J LJli. ! i Mm i I 1 i ! I i —! 1 < -L L -! ; : "" , . - —. --- • —• . ; N O I S E ; Sample of chart trace for: sensor applied to forefinger; subject female, age 24; and trace scale = 25msec/div. < = i f " • - — — - - - — _ -• -— - "WXSE — -. -- - - — <t — — SLLtiz ^> --_ _ --nfjFjri - - | i Noise Sample of chart trace for: sensor applied under thigh cuff @ 165 mmHg; subject male, age 40; and trace scale = 200msec/div. example of inverted waveform Sample of chart trace for: sensor applied under upper arm tourniquet cuff @ 200 mmHg; subject female, age 24; and trace scale = 200msec/div. motion artifact induced by intentionally moving arm Sample of chart trace for: sensor applied to thumb; subject male, age 30; and trace scale = 200msec/div. motion artifact induced by intentionally moving thumb APPENDIX D - RESULTS OF LABORATORY TESTING EXPERIMENTAL PROCEDURES Transit time relative to patient blood pressure In order to obtain quantitative PWTT data, each subject will be tested to determine their "baseline" transit times to the index finger tip of each hand, and the middle toe of each foot at their resting systolic BP as measured using the Korotkoff sound method. The subject will then perform a series of physical exercises designed to elevate their BP. Measurements will be taken at these elevated BP levels to obtain the PWTT and the associated change in transit time from baseline. PWTT measurements to the finger tips shall be made with the subject seated comfortably with the arm resting on a platform such that it is approximately horizontal. Measurements to the toes shall be made with the patient seated with the legs resting horizontally in front, supported at the feet by a platform equal in height to the seat. The experimental procedure is to follow the steps given below. (1) Apply EKG electrodes. (2) Seat subject and support limb in preparation for measurements. (3) Apply the BP cuff and Doppler probe. Where PWTT measurements are being made on the arm, the BP cuff will be applied to the opposite arm. (5) Apply the pulse detector probe to the finger or toe as required. Adjust probe positioning and amplification until a strong clean pulsatile signal is achieved as viewed on the oscilloscope. Affix probe securely using surgical tape. (6) Ensure the subject is positioned comfortably. Once the subject is seated comfortably, he/she will be asked to relax and not to change position until testing on the limb is complete. (7) Ensure that a stable valid PWTT measurement is being produced by the system. (8) Proceed to take a BP reading. Record the PWTT value measured concurrent with the first appearance of the tapping indicating systolic BP as the cuff is deflated. Record the BP. Record the estimated variance of the PWTT signal obtained. (9) Repeat (8) (10) Remove the pulse detector probe, EKG leads, BP cuff and probe as necessary to perform the exercise routine. (11) Have the subject perform exercise routine to elevate blood pressure. The routine may vary from subject to subject. Unless otherwise indicated, the subject will pedal a stationary bicycle for approximately 2 minutes. (12) immediately following exercise routine, position the patient as before and re-apply the pulse detector probe, EKG leads, BP cuff and probe. This should be done as quickly as possible to minimize the drop in BP from the elevated levels. (13) Ensure that a stable valid PWTT measurement is being produced by the system. (14) Proceed to take a BP reading. Record the PWTT value measured concurrent with the first appearance of the tapping indicating systolic BP as the cuff is deflated. Record the BP. Record the estimated variance of the PWTT signal obtained. (15) Repeat (14) until the subjects BP has returned to approximately normal level (16) Repeat steps (10) - (15) to obtain a second set of results (17) Remove the test equipment and allow the patient to rest for several minutes before continuing to test another limb if necessary. DISCUSSION OF RESULTS RESULTS AND DISCUSSION In performing the tests as described previously, several complications were encountered which introduced significant error to the results obtained. Firstly, difficulty was encountered in accurately tracking the rapidly falling systolic pressure of the subjects in the post-exercise period. After deflation of the BP cuff to hear the first K sounds which indicate the systolic BP point, it was often found that these sounds almost immediately ceased as the subject's systolic BP quickly dropped below the value just measured. Further reduction of the cuff pressure to regain the K-sounds would then locate a blood pressure significantly lower than that obtained previously. This rapid change in BP coupled with the 8 sample averaging algorithm utilized in the PWTT calculation, prevented the accurate pairing of an exact systolic BP with a PWTT value corresponding to that specific BP value. The resulting error in the BP measurements as they relate to PWTT is estimated at +/- 5 mmHg. This error is most evident at high BPs (the points at which the BP and PWTT was most rapidly changing) and decreases markedly as the BP returned to near resting levels. It was also noted that PWTT values in the period immediately following exercise never returned to the values measured prior to undertaking the exercise routine. The transit time for post-exercise BPs near previously measured resting values were always significantly (10-30 msec) shorter than those measured previously. This effect was noted to be more prevalent in the legs than the arm. This hysteresis appears to stem from the short term change in muscle tonus caused by the exercise routine. It can be expected that the compliance of the arterial path would be much higher in the pre-exercise case when the muscles are relaxed than in the post exercise period when the muscles are tensed and muscle tonus is increased. The increased rigidity of the arterial path during this period could explain a shift in the PWTT from the baseline (pre-exercise) level. Based on this hypothesis, pre-exercise PWTT values are not utilized in the following data analysis. Only post exercise measurements during which the tissue compliance can be expected to remain relatively constant are considered. It should be noted that there is still a gradual relaxing of the muscle tissue over the course of the trial measurements will change the arterial path characteristics. Some amount of "white coat syndrome" or stress induced by commencing BP measurement was also noted where patient's PWTT values on the analog output were noticed to drop slightly from previously increasing or relatively constant values upon instigation of the BP cuff inflation process. In spite of the problems encountered, the results of the trials did show a marked dependence of PWTT on subject BP. A least squared sum of errors analysis was utilized to fit a linear model to each of the eight cases recorded. The models and associated coefficients of correlation are listed on the following page and plots for each case are presented. In the case of the arm trials, the slopes determined in each case were found to be relatively constant between the cases, with a mean value of 0.63. In testing performed on legs, three trials provided highly similar slopes of an average value 0.37 while one was fit to a slope of 0.84. It should be noted that the results obtained from this subject were found to be particularly erratic due to rapidly changing blood pressure and what can be described as mild hyperactivity. In spite of the rather large errors associated with these results a strong relationship between PWTT and systolic BP appears to be evident. In addition, the general agreement of the slope values determined for the least square model indicates a good possibility for determining a set of generic slope values which could be utilized to allow modeling of a the PWTT Vs BP relationship in any patient given a single corresponding BP / PWTT pair. SUGGESTIONS FOR FURTHER WORK In order to verify the results obtained above and obtain slope information of sufficient accuracy to facilitate the development of a general algorithm for use in the control of an adaptive tourniquet it becomes necessary to adapt the experiment so as to either avoid or compensate for the problems caused by rapidly changing subject BP and hysteresis due to muscle tonus. It would be highly desirable to obtain data from subjects whose BP can be expected to shift slowly such that the an average PWTT can be determined over a sample period that is small compared to the rate of BP change. This would allow the accurate pairing of PWTT values with subject BP without the large error introduced by "chasing" rapidly shifting BP values. It would also be desirable that the shift in BP be a result of some passive means that does not result in a significant change in the muscle tonus of the subject or require interruption of the measurement process to instigate. One obvious method to overcome these problems is to perform testing on subjects who are under the influence of anesthetic in the operating room. During the course of an operation, patients typically undergo gradual BP shifts on the order of 20-30 mmHg. Data obtained from such subjects would prove would allow the accurate pairing of PWTT values measured using the prototype system with BP values as measured by the anesthesiologist. As the patient is not conscious during the measurement process it is expected that changes in muscle tonus or spontaneous shifts due to subject awareness of the measurement process will not be significant. 206 LABORATORY TESTING RESULTS FOR 4 SUBJECTS PLOTTED WITH LINEAR REGRESSION MODELS DETERMINED BY LEAST SUM OF SQUARES ANALYSIS RESULTS FOR PULSE WAVE TRANSIT TIME vs MEASURED SYSTOLIC BLOOD PRESSSURE PRELIMINARY LABORATORY TESTING ERROR IS APPROX. + 1-3 msec, +/-5mmHg SUBJECTS 1,2,3,4 - PULSE PROBE ON MIDDLE DIGIT OF RIGHT ARM 145 140 a 135 X I 130 m 125 120 115 270 CORRELATION 0.90 CASE B1 .A - RIGHT ARM • Measured O Linear Model = 309-0.6'PWTT go D • 280 290 300 PWTT (msec) 310 320 co X E _§ a. co 180 -170 160 150 140 130 120 180 CASE B2 .A - RIGHT ARM CORRELATION 0.77 • Measured D Linear Model = 282-0.58 "PWTT s • • 200 220 240 PWTT (msec) 260 280 207 155 CASE B3.A - RIGHT ARM 150 D • Measured 145 • a Linear Model = 291-0.64*PWTT "3 X 140 y E E 135 a a CORRELATION 0.94 a. m 130 125 120 1 1 R • • • • • • • 0 210 220 230 240 250 260 270 280 PWTT (msec) CASE B4.A - RIGHT ARM 165 -160 155 150 • • 0 • • • Measured a Linear Model = 0.71 "PWTT 319-mmHg) 145 140 •• CORRELATION 0.95 BP ( 135 -130 -125 120 115 D • • • • • • • • • 225 235 245 255 265 275 285 PWTT (msec) 208 SUBJECTS 1,2,3,4 -PULSE PROBE ON MIDDLE TOE OF RIGHT LEG 138 136 134 132 130 128 126 124 122 120 118 116 350 CASE B1.L = RIGHT LEG • Measured n Linear Model = 261-0.35 "PWTT CORRELATION 0.89 • • • 360 370 380 PWTT (msec) 390 400 410 cn X E E a. m 160 155 150 145 140 135 130 125 120 115 110 290 CASE B2.L - RIGHT LEG CORRELATION 0.96 • Measured Q Linear Model =409-0.84*PWTT 300 310 320 330 PWTT (msec) 340 350 360 2 0 9 CASE B3.L - RIGHT LEG CO X E E a. m 155 150 145 140 135 130 125 120 115 290 CORRELATION 0.87 • Measured O Linear Model = 258-0.38*PWTT go 300 310 320 330 PWTT (msec) 340 350 360 CASE B4.L - RIGHT LEG 155 a • Measured 150 145 • • o D Linear Model = 0.39* PWTT 268-BP (mmH 140 135 130 125 CORRELATION 0.90 • • • • • 1 120 290 300 310 320 330 340 350 360 PWTT (msec) 210 APPENDIX E - APPROVAL FORMS AND RESULTS FROM CLINICAL TESTING 211 CERTIFICATE OF APPROVAL Tha University of British Columbia Offlca of Research Services Clinical Scraenlng Committee far Research Involving Human Subjects CBrtlficate of Approval * W C 3 > A L wvgsno*roS~ McGraw. R.W. G S P A R T M E . N T Orthopaedics -~r:'C93-Q5Q7. INVnTU^eN(S) W H A A t B H i A f f C H W U . > i CARRie-3 CUT Vancouver General Hospital ccuNvcraMTCf l * McEwen, A. , Electrical Engineering S P C N « O J » N O A G E N C 2 8 National Research Council !A.a-. . . ... .. ;—: :—; Evaluation of an advancsd toumiauet svstern in surgery A ^ D * C Y A _ 2 * re SEP 1 1933 i 1 i cisTWCATC.-« The protocol and conssnt form fcr the above-named project havs been reviewed by the Committee and the experimental procedures were found to be acceptable on ethical grounds for research involving human subjects. Dr. B. McGillivrav, Chair or J L Dt/k. D.Spratley t Dr. A. Hannam, Aisociats Chair v Director, Research Services This Certificate of Approval is valid for the above term provided there is no change in the experimental procedures 212 SAMPLE RESULTS FROM INITIAL CLINICAL TRIALS (CASE BD7) CLINICAL MEASUREMENT OF CHANGES IN PU LSE W A V E 1 RANSIT TIME AS THEY RELATE TO SYSTOLIC BLOOD PRESSURE I START TIM " 1:45 PM DATA COLLECTION FORM DATE: 5/2T794 | --• -PATIENT IN FORMATIO I.D.code: B b ? sex M location: UBC age: 27 attending surgeon: BRIAN DAY procedure: KNEEARTH ROSCOPY .— • -comrnerts: —• 300 SYSTEM APPLICATION: EKG: | HONEYWEl cuff pressure: puke probe ocabon: LEFT INDEX WiGER pre-op 8P: 133/75 comments: BP measurement method: DINAMPPATIENT HON ITOR DATA mntmt JTOMATEDBP " " " " PWTT vote msec BPM DBP MBP SBP 206/212 317/327 72 66 89 114 SUBJECT TWTCHNG 208 320 68 87 118 " T41 AND CNVLSING 204/2.18 314/335 70 71 79 98 DRNGINDU ERRATIC PI CTION 204/212 314/326 59 68 87 108 JISE 21 323 65 70 85 108 206 317 54 69 85 107 1.88 290 53 63 83 gi 115 1.32/1.98 296/305 50 74 115 1.9 293 49 74 90 123 177/1.85 2737295 43 73 90 122 1.79 276 44 75 87 120 128 1.73/1.73 276/266 45 78 94 1.74/1.72 2667265 47 77 92 128 1.7/1.79 2627268 45 75 90 127 1.68/174 271/268 43 78 95 127 Pagel 213 PAGE 272 I.D.code BD7 A UTOMATED BP — — PWTT SBP" vow msec BPff " DBP MBP 1.78 274 ' 42 "76 102 127 1.7 81 102 1.72/1.68 265/259 44 84 113 148 ERRATICR USE 1.66 256 64 106 114 165 •- .- -— - — :— .. •— - - — — - ---214 APPLICATION OF ALGORITHM FOR IMPROVED APPARATUS TO DATA FROM INITIAL CLINICAL TRIALS ADAPTIVE CONTROL OF TOURNIQUET BASED ON USE OF DINIMAP AND PWTT SEE NOTES 1-9 111 I2) I3) I") [5) [6] [7] [8] re PWTT SBP SAMPLE SHIFT CORR ERROR MODE Pcuff AVG BD1 237 142 1 0 X x fixed 300 233 148 2 6 X x fixed 300 263 118 3 30 0.99 0 PWTT 220 273/220 BD2 330 96 1 0 X x fixed 300 311 103 2 7 X x fixed 300 314 102 3 7 X x fixed 300 302 110 4 14 0.97 x fixed 300 317 103 5 14 0.96 -2 PWTT 190 ' 308 104 6 14 0.95 1 PWTT 200 • 312 111 7 15 0.8 -9 PWTT 210 * 308 102 8 15 0.73 4 fixed 300 303 109 9 15 0.77 -4 fixed 300 267/200 BD3 242 226 245 276 121 129 124 115 0 8 8 14 x 0.95 x fixed x fixed x fixed x fixed 300 300 300 300 300/-BD4 291 339 305 325 333 336 276 111 90 98 98 92 93 110 0 21 21 21 21 21 21 x 1 0.93 0.91 0.92 0.93 0.94 x fixed x fixed 6 PWTT -4 PWTT 1 PWTT -3 PWTT 5 PWTT 300 300 190 ' 190 • 190 • 190 ' 210 ' 224/194 BD5 242 273 273 285 264 253 262 142 112 121 113 115 123 115 0 30 30 30 30 30 30 x 1 0.96 0.95 0.89 0.86 0.82 x fixed x fixed -10 fixed -6 PWTT 10 fixed 8 fixed 7 PWTT 300 300 300 210 ' 300 300 210 ' 274/210 Page 1 SEE NOTES 1-9 CASE BD7 [11 [21 |3] ["1 [51 [61 [71 [8] WTT SBP SAMPLE SHIFT CORR ERROR MODE Pcuff A 320 108 1 0 X X (ixed 300 323 108 2 1 X X fixed 300 317 107 3 8 X X fixed 300 290 115 4 8 X X fixed 300 300 115 5 15 X X fixed 300 279 122 6 15 X X fixed 300 276 120 7 21 0.97 6 PWTT 220 • 271 128 8 21 0.96 -6 PWTT 230 * 266 128 9 21 0.97 -4 PWTT 230 * 268 127 10 21 0.97 -1 PWTT 230 • 269 127 11 21 0.97 -1 PWTT 230 • 274 127 12 21 0.97 -3 PWTT 230 • 262 135 13 28 0.96 -7 PWTT 250 ' [91 263/231 BD8 306 110 1 0 x x fixed 300 3 1 2 108 2 2 x x fixed 300 3 1 8 108 3 2 x x fixed 300 3 2 9 99 4 11 x x fixed 300 3 2 8 100 5 11 0.93 0 PWTT 190* 32<1 109 6 11 0.95 -6 PWTT 200* 3 0 5 109 7 11 0.79 1 fixed 300 270/195 BD9 341 83 1 0 x x fixed 300 347 100 2 17 1 x fixed 300 299 100 3 17 0 4 x fixed 300 314 121 4 38 048 -25 fixed 300 289 118 5 38 0.63 -6 fixed 300 280 117 6 38 0.69 0 fixed 300 278 119 7 38 0.73 -1 fixed 300 288 123 8 40 0 75 -7 fixed 300 276 130 9 47 078 -8 fixed 300 277 121 10 47 0.79 -2 fixed 300 254 129 11 47 0.81 2 PWTT 240 267 131 12 48 0.83 -5 PWTT 240 261 136 13 53 0 85 -7 PWTT 250 ' 234 133 14 53 0 85 10 fixed 300 241 132 15 53 0 85 6 fixed 300 243 141 16 58 0.85 -6 PWTT 260 287/248 Page 2 216 [1! CASE PWTT bd10 265 279 286 296 304 319 311 303 bd11 313 306 280 SEE NOTES 1-9 I2] [3] [41 PI (6) [7] [81 SBP SAMPLE SHIFT CORR ERROR MODE Pcuff 125 1 0 X x (ixed 300 111 2 14 1 x (ixed 300 97 3 28 0.98 7 PWTT 190 94 4 31 0.97 -10 fixed 300 91 5 34 • 0.96 -8 fixed 300 88 6 37 092 -14 fixed 300 94 7 37 0.91 -7 fixed 300 108 8 37 0.9 -15 fixed 300 96 1 0 X x fixed 300 107 2 11 1 x fixed 300 107 3 11 0.66 x fixed 300 [91 AVG 286/190 300/ -TOTAL TIME WEIGHTED AVERAGE 273/216 NOTES: [1] PWTT in msec as measured using experimental system [2] Systolic blood pressure in mmHg as measured by OR. monitor "" Sample number in order from start of procedure.samples at 4 minute intervals Maximum shift in SBP since beginning of measurement (no model until at least 10) Coefficient of correlation of LR model formed using all data to current sample point, x if no model Error in using last LR model to estimate current BP as determined using O R monitor, x if no mod< [7] Mode of control based on guidelines listed in text Cuff pressure in mmHg. set at 300 for fixed mode or (SBP) "1.6 + 25 in PWTT mode All values rounded up in 10 mmHg intervals Time averaged cuff pressure in mmHg. Total average /adaptive mode average. PI (41 [51 161 [81 [91 Page 3 217 SAMPLE OF DATA AND ANALYSIS FROM PROSPECTIVE STUDY (CASEBD2.3) CLINICAL TESTING Ol- IMPROVED LOP ESTIMATION APPARATUS DATA COLLECTION F O R M PATIENT INFORMATION: LD.coda sex age: procedure: commenls: BDTJ" W 37 STAPLE REMOVAL TJATST TIMTf Tjrauite) ~1 "IT "~T2l 15 UL 17 m-.— pulse probe location: comments: volts PWTT [msec "7.84 ~T.S 3 1 1.8 IM J.78 1.76 1:92 ~T.9 ~T.1 "19T " T T J ~7.84 ~im "1.74 283 _ ??3 "293 __277 285 271 3296 293 177 194 274 '283 157 283 159 LEFT THUMB MBP 75 "73 84! "38 SBP 104: 111 116 127 DATE:"" location: UBC 0CT24" surgeon: SH/Fr 12 START TIME DR.DAY 10:40 cuff pressure: pre-op BP: "300 MODEL 313-72 308-.71 CORR XERROR 0.7 0.93 Pagel 218 TIME PW1 T BP HHP""" SBP SMF[_ "" MODEL 317-73 (minute) vofts msec 93 10 1.1 277 19 1.78 274 ' 124 20 1.77 m 21 22 1.6 246 23 1.7 262 24 1.7 262 320-74 25 1.74 268 89 122 26 1.75 270 256 27 1.66 28 1.68 259 29 1.72 265 30 1.6 246 93 128 266-.55 31 1.6 246 274-57 32 1.6 246 . 33 1.62 249 34 1.7 262 35 1.56 240 104 36 1.63 25T 246 37 1.6 38 1.68 259 39 1.78 274 40 41 42 : - • - - -_ — 43 44 45 46 47 ; 48 49 50 5T 52 53 54 55 56 — 57 — 58 59 60 \CORR \EMffi "0.87 1.861 0.91 10 1 0 Page 2 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0065135/manifest

Comment

Related Items