UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Head acceleration during balance Roskell, Melanie 2008

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


ubc_2009_spring_roskell_melanie.pdf [ 1.73MB ]
JSON: 1.0067055.json
JSON-LD: 1.0067055+ld.json
RDF/XML (Pretty): 1.0067055.xml
RDF/JSON: 1.0067055+rdf.json
Turtle: 1.0067055+rdf-turtle.txt
N-Triples: 1.0067055+rdf-ntriples.txt
Original Record: 1.0067055 +original-record.json
Full Text

Full Text

//HEAD ACCELERATION DURING BALANCEbyMELANIE ROSKELLB.A.Sc, University of Waterloo, 2006A THESIS SUBMIHED ll”J PARTIAL FULFILLMENT OFTHE REQUIREMENTS FOR THE DEGREE OFMASTER OF SCIENCEinTHE FACULTY OF GRADUATE STUDIES(Human Kinetics)The University of British Columbia(Vancouver)November 2008© Melanie RoskeLl, 2008AbstractThe overall purpose of this thesis was to study the angular and linear accelerations thatoccur at the head during quiet standing in healthy humans. To date, there were few descriptionsof linear head accelerations in quiet standing, and no focus on angular head accelerations. Thecontribution of the vestibular system to standing balance can be better understood fromrecognizing these, the stimuli that the system experiences during the task.Head accelerations were measured under four manipulations of sensory condition, andRMS and median frequency values were reported for linear and angular head accelerations inReid’s planes. Coherence was also calculated between force plate forces and headaccelerations, and between lower leg EMG and angular head accelerations in the directions ofthe semi-circular canals. This study considered two factors in the manipulation of quietstanding sway: vision (eyes open/closed) and surface (hard/compliant foam).The results show that angular head accelerations are repeatable under full sensoryconditions, and that angular head acceleration RMS is above known vestibular thresholds in alltested sensory conditions. Linear head acceleration absolute maximum and RMS valuesmatched previous reports under similar conditions. Significant coherence was found below7Hz in both coherence analyses, likely due to the mechanical linkage. This coherence alsoshowed defined troughs in varying regions, which were attributed to the interference of activesystems (visual, somatosensoiy and vestibular) on the mechanical propagation of forces. Theresults also reinforced that the inverted pendulum model is valid in quiet standing on a hardsurface in the sagittal and frontal planes.This study shows that the vestibular system is able to detect sway at the head duringquiet standing under all four sensory conditions tested. Consequently, the vestibular systemmay play a range of roles in quiet standing, which may change as its relative importance in11balance increases. The measurement of head accelerations is confirmed as a useful technique instudying balance in quietly standing humans.111Table of contentsAbstract iiTable of contents ivList of tables viList of figures viiList of equations viiiAcknowledgements ixLiterature review 1Balance and sway 1Factors affecting sway 2Anatomy and neurophysiology of the vestibular system 6Semi-circular canals 6Otolith organs 9Sensitivity 10Head acceleration during quiet stance 11Measuring head acceleration 13Vestibular innervation 14Afferent innervation 14Efferent innervation 15Reticulo- and vestibulo-spinal pathways 16Vestibulomotor reflex 18Galvanic vestibular stimulation 19Stochastic vestibular stimulation 20Research motivation 22Aims 24MethodsParticipants 25Apparatus 25Experimental procedure 27Data analysis 29Statistics 33ivResults .35Control trials 35Sensory condition trials 35Angular acceleration 36Linear acceleration 39Canal-transformed accelerations 39Coherence 40Force plate 40EMG 45Discussion 49Repeatability 49Comparison of data to previous accelerometry work 49Comparison of acceleration data to reported vestibular thresholds 50The inverted pendulum 51Purposeful sway 53Active systems in standing balance 55The role of the vestibular system in standing balance 56Limitations of the study 59Conclusions 64List of references 66Appendix 1: Tables of control trial data 74Appendix 2: Differences in acceleration input to each SCC 76Appendix 3: Coherence between head accelerations and EMG data 77Appendix 4: Center of pressure measures 82Appendix 5: Research Ethics Board Certificate of Approval 83VList of tablesTable 1: List of experimental conditions 27Table 2: Angular acceleration characteristics in roll, pitch and yaw under four sensoryconditions 37Table 3: Linear acceleration characteristics in anteroposterior (AP), mediolateral (ML) andinferiosuperior (IS) direction under four sensory conditions 39Table 4: Angular head acceleration characteristics in the directions of the canals under foursensory conditions 40Table 5: Results for angular accleration of the head in roll, pitch and yaw during three controltrials in the eyes-open, hard-surface condition 74Table 6: Results for linear accieration of the head in the anteroposterior (AP), mediolateral(ML) and inferiosuperior (IS) directions during three control trials in the eyes-open,hard-surface condition 75Table 7: Results for COP displacement of in the anteroposterior (AP) and mediolateral (ML)directions during three control trials in the EOHS condition 82Table 8: Results for COP displacement of in the anteroposterior (AP) and mediolateral (ML)directions during two trials: EOHS and ECHS 82viList of figuresFigure 1: Setup of the accelerometer array on a subject’s head 28Figure 2: Angular acceleration RMS under four sensory conditions in three directions 38Figure 3: Angular acceleration median frequency in three directions 38Figure 4: Coherence and TCD plots for Reid head accelerations vs. force plate forces 44Figure 5: Head accelerations in the direction of the SCCs (anterior, posterior and horizontalcanals as marked) vs. EMG from bilateral soleus 47Figure 6: TCD plots between anterior, posterior and horizontal semi-circular canalaccelerations and soleus EMG under four sensory conditions 48Figure 7: Angular and linear accelerations about Reid’s axes for one EOHS trial in arepresentative subject over 4.5 minutes 61Figure 8: Processed accelerometer data from one second in the middle of one EOHS trial in arepresentative subject 61Figure 9: Coherence between anterior canal accelerations and EMG 78Figure 10: Coherence between right horizontal canal accelerations and EMG 79Figure 11: Coherence between posterior canal accelerations and EMG 80Figure 12: TCD plots between canal accelerations and EMG under four sensory conditions . . .81viiList of equationsEquation 1: Angular acceleration equation for the head, where: ai is angular acceleration aboutaxis i, Ai is linear acceleration from accelerometer i, and di is the distance betweenaccelerometers 30Equation 2: Coherence between signals A and B is calculated by dividing the cross-spectra’smagnitude squared, by the product of the two autospectra 32viiiAcknowledgementsI would like to thank my supervisor, Dr. Timothy Inglis, for his guidance and support.Thank you so much for teaching me and having confidence in me. I would like to thank mycommittee, Dr. Jean-Sebastien Blouin and Dr. Mark Carpenter, for their advice and help in somany areas. Thank you all for backing this thesis so enthusiastically; your expertise and energyhave been invaluable. I would also like to thank my fellow graduate students, all willingsubjects, for their encouragement and camaraderie. A special thank you goes to my colleaguesin the Human Kinetics labs, especially the Human and Spine Neurophysiology labs, who havemade this process extra colourful. Thank you for growing with me. Lastly I would like to thankmy family, for their never-ending supply of support and good vibes.ixLiterature reviewBalance and swayBalance is the critical task of preserving equilibrium of the body segments with respect togravity. This task requires maintenance of the body’s center of mass within the base of support,although stabilization of the trunk and head in space may take priority during balance (Horak &MacPherson, 1996; Latt et al., 2008). Muscle stiffness and tonic activity, including paradoxicalmuscle contractions, can account for a proportion of the successful execution of balance, as canpassive joint stiffness and long-latency functional stretch reflexes (Horak & MacPherson, 1996;Nashner, 1976). Balance is no simple task, however, and it requires the convergence of manytypes of sensory information to achieve. The everyday posture of quiet standing uses input fromthree main sensory systems: somatosensory, visual and vestibular. The many channels ofavailable sensory information lead to redundancies, which help in the interpretation anddisambiguation of each sense’s output. Exactly how this information is used to accomplishbalance is not fully understood, but it is known to be dependent on task and context (Horak &MacPherson, 1996).It is well known that people sway while standing quietly, and that the amount and qualityof information available from these three sensory systems affects that sway (as do other factors).Sway during quiet standing in healthy humans is usually described as an inverted pendulum,with responses to regular sway controlled by rotations around the ankles (Horak & MacPherson,1996; Winter, 1995). Within this inverted pendulum model, healthy humans with all sensoryinformation available sway maximally at about 0.3 Hz as measured at the center of mass (CoM,displacement in both mediolateral and anteroposterior directions). In power spectral density1estimates, studies have found most CoM sway power below 1Hz (Jeka et al., 2004), althoughsome power up to 5Hz has been reported (summarized in Winter, 1995). Power spectral densityestimates of center of pressure (CoP) data have similar results: 90% of power is below 2Hz, withmaximums between 0.30-1.20 Hz in the anteroposterior direction, and 0.30-0.90 Hz in themediolateral direction (Soames & Atha, 1982). These results were based on sampling durationsof 4 minutes, and therefore were long enough to catch low-frequency components of swayaccording to Carpenter et al. (2001).Factors affecting swaySway during quiet stance is greatly affected by a number of factors, and thus can beaffected by eliminating or producing errors in the signals from the three sensory systems thatmaintain standing balance: somatosensory, visual and vestibular.Somatosensory contributions to balance (including skin touch, pressure andproprioception) is often measured with patients with peripheral neuropathy (Lafond et a]., 2004;Simoneau et al., 1994), or in healthy subjects by providing erroneous somatosensoryinformation. This erroneous signal can be achieved by having the subject stand on a sway-referenced platform (as in Nashner, 1971) or a block of thick foam (Horak & MacPherson, 1996;Jeka et al., 2004). Both of these methods limit the usefulness of the somatosensory afference,mostly from the feet and ankles. These methods have shown that sway amplitude as measuredfrom the center of pressure or center of mass increases when subjects have less somatosensoryinformation available to them (Dietz, 1992; Horak & MacPherson, 1996; Jeka et al., 2004). Swaycan be reduced by introducing further somatosensory information, for example, by touchingsomething stationary and external with another part of the body. This extra information decreases2sway even when vision is present (Jeka & Lackner, 1994). Touching one’s own body can alsodecrease sway (Nagano et al., 2006).Visual information is also used to stabilize sway in standing. People with their eyesclosed sway 30% more that they do with their eyes open in a feet-apart stance (as reported byRomberg, reviewed in Horak & MacPherson, 1996). It has also been shown that the variability ofpostural sway is decreased when the eyes are engaged in a tracking task (Stoffregen et a!., 2007).Vision is thought to contribute most to standing balance in frequencies below 1Hz (Diener &Dichgans, 1988).The vestibular system is likely the least understood of the sensory systems that contributeto balance. Patients with vestibular lesions are often tested to try to decipher the importance ofvestibular input to the maintenance of quiet stance, although erroneous signals can also beintroduced to a healthy population by electrical stimulation. Patients with bilateral vestibular losscan exhibit sway within the range of normals when under full sensory conditions (Dietz, 1992;Yoneda & Tokumasu, 1986). They are usually able to stand without one of the two remainingbalance systems (somatosensory and vision), but without both they will fall when attempting tostand quietly (Nashner et al., 1982). This is strong evidence for some role for vestibular inputduring quiet standing, yet Fitzpatrick and McCloskey (1994) found that healthy people cannotconsciously perceive vestibular inputs while standing quietly, while somatosensory(proprioceptive) and visual inputs were consciously perceived during normal sway. The authorsnoted that subconscious processing of vestibular inputs is likely. Indeed, in 1982 Nashner, Blackand Wall (Nashner et al., 1982) suggested that vestibular information was not used consciouslyas an indicator of instability. They proposed that vestibular input was processed subconsciouslyat a high level, and used as an internal reference by which to resolve conflicts in visual and3somatosensory information. This view of the vestibular system as more of a ‘quiet partner’ instanding balance is also supported by Day and Fitzpatrick, who call the vestibular system the“silent sense” in balance (Day & Fitzpatrick, 2005). It also may be supported by the presence ofa whole-body balance response to selective electrical stimulation of the vestibular organs(Fitzpatrick & Day, 2004), for if the vestibular system created responses without other sensoryinput, one might expect the response to be confined to the head.It may appear that the vestibular system is a prime candidate for an internal reference ofthe vertical (the direction of the gravity vector). This hypothesis is further supported by studiesthat show illusory sway in supported subjects presented with artificial vestibular signals (Horak& Macpherson, 1996). However, experiments conducted in an environment lacking much usefulsomatosensory information (in this case, immersion in a tank of water) found that estimates ofthe direction of the vertical under these vestibular-dependent conditions were highly erroneouscompared to a land-based condition with intact somatosensory sense (+1- 20 degrees vs. +1- 1-2degrees; Nelson, 1968; Horak & Macpherson, 1996). It appears, therefore, that the vestibularsystem alone does not account for an internal reference to the vertical.Another proposed main use for vestibular contributions in standing is active stabilizationof the head in space (Horak & Macpherson, 1996). It has been shown that patients with completevestibular loss refrain from using a hip strategy when perturbed (which is the situation in which acontrol subject would likely use a hip strategy; recall that the ankle strategy is principal in quietstanding). The hip strategy requires a counterrotation of the head and trunk, and tends to stabilizethe head to the environment. The reluctance of vestibular patients to adopt this strategy seems toshow that they have difficulty with active head stabilization (Horak & Macpherson, 1996; Blacket al., 1988), perhaps due to their vestibular lesion. As reviewed in the following sections, the4vestibular system is designed to detect head motion, and therefore seems a most suitablecandidate for the active stabilization of the head in space.The vestibular system may also contribute to standing balance through the execution ofthe vestibulospinal reflex. This reflex as evoked by electrical stimulation of the vestibular nervehas been shown to increase in magnitude in the lower leg muscles under ‘vestibular dependent’standing conditions (for example, eyes closed on a compliant surface; Welgampola & Colebatch,2001). Welgampola and Colebatch (2001) thought that this might point to the vestibular systemas a “backup” sensory system, intended to maintain balance when vision and somatosensorysenses have “failed”.It may be that the vestibular system in quiet standing is used for all or none of: resolvingalternative sensory conflict; an internal reference of the vertical; active head stabilization; or theamplitude modulation and release of the vestibulospinal reflex. There are of course otherpossibilities, as much remains unknown about the normal uses of this system in quiet stance. Oneof the reasons for this may be the way that this system has mainly been studied in the past.As previously mentioned, the vestibular contributions to standing in humans have mostlybeen inferred from studies using vestibular lesion patients. However, these patients have almostcertainly been afflicted with their condition for long periods of time and may have alreadyadapted to their deficit, using other sensory systems in compensatory strategies that are not wellunderstood. Indeed, studies in cats show that otherwise healthy animals who have beenbilaterally lesioned have substantial deficits in balance (including an inability to stand at all)immediately following the lesion (Macpherson & Inglis, 1993). These animals appeared to adaptrapidly and by the end of the first month had regained the ability to stand and walk stably.Studies using vestibular lesion patients, then, include a confounding factor in regards to the5compensation strategies that adapted lesion patients develop over time. Therefore, it remainsunclear how vestibular information is used during standing sway in healthy humans underdifferent sensory conditions. A closer look at the vestibular system’s location, anatomy andneurophysiology may aid in deciding how this could best be explored.Anatomy and neurophysiology of the vestibular systemThe human vestibular system is located in the inner ear on both sides of the head, rigidlyembedded in the skull. It is a continuous bony structure lined with membrane and filled withfluid. The system consists of two different types of structures: the three semi-circular canals andthe two otolith organs.Semi-circular canalsLocation and orientationThe semi-circular canals (SCCs) are three ducts located within bony loop-shaped cavitiesembedded in the inner ear, within the petrous part of the temporal bone (Drake et al., 2005).These canals are approximately mutually orthogonal and are called the horizontal (or lateral),posterior, and anterior (or superior) canals. The SCCs are generally considered responsible forthe detection of angular acceleration of the head in their respective planes.A recent study involving CT scans of human skulls revealed detailed information on theorientations of the SCCs within the skull (Della Santina et al., 2005). The researchers usedreference planes that are commonly used to landmark skull locations, called Reid’s planes. Todefine these planes, the interaural axis was first located. This is the axis connecting the points atthe center of each bony external auditory canal. Zpj was thus defined as the plane thatcontained the interaural axis and the points on the edge of the most inferior part of each6infraorbital rim (this is also known as the Frankfort plane; Drake et a!., 2005). The plane wasover defined, but the study found all points to lie within 0.5mm of it. YpJD was defined as theplane bisecting and perpendicular to the interaural axis, and X1ID was the plane perpendicular toZpjJD and containing the interaural axis. The origin was located at the center of the left externalauditory canal. The orientations of the canals were found in relation to these reference planes.The horizontal canal was angled upwards towards the eyes at approximately 20 degrees to ZJD.The posterior and anterior canals were both perpendicular to the horizontal canal, and were atapproximately 48.5 (+1- 5.1) degrees and 38.4 (+1- 5.1) degrees to the YD plane, in opposingdirections. The SCCs were found to be very close to being mutually orthogonal, but were not.This means that rotation about any axis would inevitably stimulate more than one SCC at a time.Reid’s axes are used in this thesis, yet it is important to note that the direction conventionhas been changed. In the Reid’s axes used in this document, XpJD remains positive forward butYID is now positive right, and ZpJD is positive down. All reported results use this newdefinition of Reid’s axes.Each of the canals has a mirrored twin in the vestibular apparatus on the opposite side ofthe head. The orientation difference between these is negligible (Della Santina et al., 2005),meaning that the twin canal provides an inverse duplicate signal. This makes one set of canalseffectively redundant. This arrangement may decrease error in vestibular conclusions byimproving amplitude and directional sensitivity (Fitzpatrick & Day, 2004).The SCCs in most mammals are approximately equal in size and are proportionallyrelated to the size of the mammal. In humans, however, the horizontal canal is smaller than theother two (Spoor et al., 1994). This may be because the anterior and posterior canals togethercode for angular acceleration in the sagittal and frontal planes, which are most important for the7maintenance of balance (Fitzpatrick et al., 2006). Evolution may have therefore favoured largeranterior and posterior canals for bipedal stance (Day & Fitzpatrick, 2005). Larger canals aremore sensitive to smaller perturbations, meaning that the horizontal SCC may be physiologicallyless sensitive than the other two. It appears, however, that rotation in its axis is perceived better(see section “Sensitivity”, page 10).PhysiologyThe bony canals are filled with perilymph fluid and lined with a membrane which is itselffilled with endolymph fluid. Each SCC contains a bulging area near its base called the ampulla,which contains the sensory transducer called the cupula. The cupula consists of many hairs, thestereocilia, and one large hair on the outer edge called the kinocilium. Head acceleration causesthe bony canals to move with the head, and the endolymph fluid lags due to inertia and theviscous properties of the fluid. This lag causes a pressure differential which pushes the cupula inthe ampullae, causing the hairs inside to bend either towards or away from the kinocilium. Thiscauses an increase or a decrease (respectively) in the firing rate of the vestibular nerve endingattached to the cupula (Fitzpatrick & Day, 2004).Although the SCCs are generally considered to be responsible for detecting angularacceleration, the viscous properties of the endolymph fluid in the ducts actually causes the SCCsto act as angular velocity transducers under a certain band of frequencies (Highstein et al., 2005;Lysakowski & Goldberg, 2004). Mathematically, this band is approximately 0.025 — 30 Hz. Atfrequencies below 0.025Hz, the stiffness of the cupula and mass of the endolymph fluid causethe SCCs to act as acceleration transducers directly. At frequencies over 3 0Hz, the same factorscause the canals to act as displacement transducers. This temporal filtering is a characteristic ofthe fluid movement in the canals, occurring before mechanotransduction from the hair cells8(Rabbitt et a!., 2004). Some collateral nerve fibres between hair cells tend to act as differentiatorsbetween the hair cells and the vestibular nuclei (Ross, 2003), meaning that at least someacceleration information is still conveyed to those nuclei even when in the velocity-transducingfrequency range.Otolith organsLocation and orientationThere are two otolith organs in the inner ear. They are called the utricle and saccule, andthey are responsible for detecting linear acceleration in the approximate horizontal and verticalplanes, respectively. The otoliths are membranous sacs located in a cavity called the vestibule,inferior to the SCCs. The utricle is attached to all three of the SCCs, and they drain endolymphfluid into it on the side opposite to the ampullae. The saccule also shares endolymph fluidsthrough a small passage called the utricosaccular duct (Drake et al., 2005). In this way, all of thevestibular organs share common endolymph fluids.PhysiologyThe otoliths each contain a sensory section of membrane called the macula (or macule).The macula is approximately flat, and is covered in 20,000-30,000 stereocilia (Fitzpatrick &Day, 2004). There are also many kinocilia, spread across the surface of the maculae with theirassociated stereocilia. All of the hair receptors on the macula are embedded in a gel-like fibroussubstance, which is topped with otoconia. The otoconia are crystals of calcium carbonateattached to the top of the gel, and they provide inertia to the system. When the head isaccelerated the otoconia lag behind, bending the gel and embedded hairs in the direction oppositethe acceleration (Blumenfeld, 2002). This action bends the stereocilia towards or away from theirkinocilia, and changes the firing rate of the attached nerve rootlets.9The directional sensitivity of each bundle of hair receptors varies across the macula,forming a curved line of maximal sensitivity near the middle of the macula. This line is calledthe striola, and it is the movement of the stereocilia towards or away from the striola whichcauses an increase or decrease in the firing rate of the attached nerves. The striola is curved inorder to provide information in many directions in the plane. The hair cells in the saccularmacula are arranged to have a directional sensitivity away from the striola, whereas the utricularmacula’s polarity is towards the striola (Fitzpatrick & Day, 2004).The morphology of the otolith organs makes them sensitive to linear accelerations, whichincludes gravity according to Einstein’s equivalence principle. Unfortunately this leads to anambiguity in otolith output, which becomes problematic mostly at high frequencies (Nashner,1971). This ambiguity may be solved by input from the SCCs, as shown from vestibular nucleusrecordings in monkeys (Angelaki et al., 2004) and cats (Uchino et al., 2005).SensitivityThe vestibular organs certainly seem ideal for detecting sway at the head during quietstance. Meiry (1966) found otolith sensitivity to be approximately 0.059 m/s2 (6 milli-g).However, Nashner reasoned that the otolith organs likely play no role in detecting body sway(1971). According to mathematical models, otolith sensitivity is highest at a very low frequencyof body sway and therefore its function is most likely relegated to encoding a static verticalreference. However the SCCs, in the pitch axis at least, have a calculated threshold ofapproximately 0.05 deg/sec2and are sensitive enough to detect the small accelerations that occurat the head during sway in quiet stance (according to Nashner, 1971). They are likely mosteffective in coding information at head sway frequencies above 0.1Hz (Nashner et al., 1989),10which is in the lower range of the head sway power reported by Easton et al. (0.15Hz; Easton etal., 1998).In 1989, a group led by Nashner plotted theoretical SCC and otolith thresholds, alongwith theoretical limits on the use of a pure ankle strategy, to discover whether the vestibularorgans could detect the accelerations that occur at the head during quiet standing sway. Twoareas were labeled to represent typical values of sway for vestibular-only and full sensoryconditions. It appeared that the vestibular organs were theoretically able to detect headaccelerations in quiet stance under vestibular-dependent sensory conditions. This was atheoretical model, however, and its conclusions under full sensory conditions were uncertain (seeNashner et al., 1989).The perceptual threshold of the SCCs has been experimentally measured, and in velocityterms is about 1.5 deg/s about the vertical axis (stimulating the horizontal canal; Benson et al.,1989). Thresholds about the two axes in the horizontal plane were discovered to be higher, at2.04 deg/s in roll and 2.07 deg/s in pitch. A previous study using acceleration as the stimulusfound no significant difference between the SCC thresholds (Clark & Stewart, 1970). Thesestudies were completed using turning tables or seats, however. It is still unclear whether thesetheoretical and measured thresholds are exceeded by the motions of the head (to which thevestibular apparatus is rigidly attached) during normal standing sway.Head acceleration during quiet stanceMost of the balance studies mentioned in the preceding sections used CoP as measuredfrom a force plate to infer stability of the participants. Since my thesis is focused on stability atthe head, it is important to note that a comparison of CoP and head sway measures found thatthey infer stability in a similar manner (Sakaguchi et al., 1995). A correlation analysis using11linear acceleration data found a very strong (“almost one”) correlation between head sway andcenter of gravity sway in equivalent directions (i.e. both head and center of gravity inmediolateral direction; Matsubara et al., 1983; Miyoshi et al., 1983). Therefore, it may not beunreasonable to expect head acceleration results to follow similar trends to those discussed in thepreceding sections (which are mostly CoP results). There are some studies, however, that havemeasured head acceleration directly, although not so thoroughly as the CoP studies.There is little reporting of angular head accelerations during quiet stance in the literature,but a technique called “acceleration registrography” has been used to record linear headaccelerations along one axis (Kitahara, 1965; Tsuj ikawa, 1966). One study using this techniqueto measure mediolateral sway showed head accelerations of approximately 20.8 milli-g (in the10-second period with the highest accelerations) during quiet standing in a normal stance, withno difference when vision was excluded (Tsujikawa, 1966). In a tandem stance, theseaccelerations increased to a maximum of 25.5 milli-g with the eyes open, and 45.5 milli-g withthe eyes closed. In both stances, brief periods of acceleration up to 250 milli-g were observed. Amore recent study by Winter et al. (1998) calculated linear head acceleration from headdisplacement data during quiet standing trials, with the eyes open in a comfortable stance. Theyfound mean linear head accelerations from 0.016 - 0.018 m/s2 (—1.7 milli-g) in theanteroposterior direction, and 0.01 — 0.012 (—‘1.1 milli-g) mediolaterally.In even more recent work, head movement information is sporadically mentioned as acontrol condition, usually in comparison to head movement profiles from platform translation ormoving visual fields (Easton et al., 1998; Kelly et al., 2005; Keshner & Dhaher, 2008).Frequency analyses on head sway data during quiet stance are scarce. One study, focused on theeffectiveness of sound as a balance aid, reported as a control condition a maximum mean power12of 0.15Hz for the displacement of the head in the mediolateral direction during quiet stance inhealthy subjects with eyes closed (Easton et al., 1998). Power was reported from 0.05Hz to0.58Hz overall.Information on linear head accelerations during quiet stance are limited in currentliterature; those on angular accelerations even more so. A description of the accelerationbehaviour of the head during quiet stance did not exist prior to this thesis. This description is ofparticular interest because the vestibular system is rigidly connected to the head, and is thereforedirectly affected by accelerations there. The vestibular system is theoretically sensitive enough todetect head accelerations, yet these accelerations had not been characterized. This may be due toonly recent developments in instrumentation sensitivity and computing power making this kindof description accurate. A complete description of linear and angular head accelerations inrelation to Reid’s axes of the head was the first objective of this thesis study.Measuring head accelerationAccurate measurement of the acceleration of the head was a fundamental part of thesuccess of this thesis. For years, it has been common practice in automobile crash testing tocalculate angular acceleration by using an array of nine linear accelerometers (Padgaonkar et al.,1975; King, 1993; Blouin et al., 2007; Yoganandan et al., 2006). The most common arrangementused is called a 3-2-2-2, and consists of the 9 accelerometers rigidly mounted on a triangularpyramidal metal frame. Three accelerometers are orthogonally arranged at the peak of thepyramid, and the other 6 are in right-angle pairs at each of the triangular base points.Theoretically, only 5 linear accelerometers are necessary to resolve the 3 angular accelerationequations; a6this required for concurrent linear acceleration calculations. In reality, the inherentnoisiness of each accelerometer may lead to an accumulation of errors, resulting in an overall13inaccurate calculation. Adding 3 more accelerometers introduces redundancy into the system,removing the opportunity for multiplication of errors and ensuring a more accurate and stableresult (Padgaonkar et al., 1975).Vestibular innervationOnce accelerations at the head during standing sway have been accurately measured anddescribed, the accelerations that the SCCs experience can also be described (since they arerigidly embedded in the skull in known planes). What effect can these stimulated vestibularorgans have on the body? A closer look at the innervation of the vestibular system and theassociated nervous connections is appropriate.Afferent innervationThe nerve that conveys all information from the vestibular organs to the brain is theeighth cranial nerve, known as the vestibulocochlear nerve. This nerve begins as either calyx orbouton endings attached to receptor hair cells in the crista of the SCCs, and the maculae of theotolith organs (Lysakowski et al., 1995; Ross, 2003). These nerve rootlets are involved in veryearly parallel processing (likely differentiation; Ross, 2003) by way of collateral inputs betweenhair cells. The rootlets, or primary vestibular afferents, then synapse in nearby ganglia beforeconverging with nerves from the neighbouring cochlea to form the aptly namedvestibulocochlear nerve. This nerve passes through the internal acoustic meatus and into the skull(Drake et al., 2005), traveling to the vestibular nuclei in the brainstem (Blumenfeld, 2002). Thesecondary vestibular nuclei reside in the brainstem, specifically in the pons and medulla (see“Reticulo- and vestibulo-spinal pathways”, page 16, for further information on the vestibularnuclei).14An important feature of the primary vestibular afferents is that they have a resting firingrate, which varies between species. It has not been directly measured in humans, but is known tobe 90-115 Hz in macaque monkeys (Cullen & Minor, 2002). This spontaneous dischargerequires no stimulus and is hypothesized to contribute to muscle tone (Lysakowski & Goldberg,2004). Its main advantage is that it allows bidirectional signals to emerge from the vestibularorgans as either an increase or a decrease in the base firing rate. The change in rate is not linear,however, as the afferents appear to be more sensitive to excitatory rather than inhibitory inputs(Ross, 2003).The type of base firing rate splits the primary vestibular afferents into two main groups:regularly and irregularly firing. Regular primary afferents make up about 75% of the totalnumber of primary afferents, with irregular making up the last 25%. There are no exact criteriafor classification, however, and sometimes “intermediate” afferents are mentioned in literature(Plotnik et al., 2005). Irregular units are associated with larger diameter primary afferents, andgenerally have a lower tonic rate, higher sensitivity and lower refractory period than regularafferents (Fitzpatrick & Day, 2004; Goldberg et al., 1984; Lysakowski & Goldberg, 2004). Theyare also more sensitive to electrical stimulation (see “Galvanic vestibular stimulation”). Theirregularity of these afferents does not appear to help code information for velocity any betterthan regularly firing afferents (Highstein et a!., 2005).Efferent innervationThe vestibular system is also equipped with an efferent nervous system, although itsfunction is still under discussion. The efferent nerves start in the brain stem, predominantly in thelateral vestibular nucleus, and innervate the vestibular system both ipsi- and contra-laterally(Klinke & Galley, 1974). In mammals, it appears that vestibular efferent excitation excites15vestibular afferents (both SCC and otolith; Plotnik et al., 2005). The efferents can be excited byrotations in any direction, but require large angular velocities for this effect to be seen. Eventhen, the excitation from efferents is much smaller than usual afferent responses. Vestibularefferents can be stimulated in other ways, as well, including direct manipulation of the otolithsand passive limb movement (as seen in rabbits and frogs, reviewed in Klinke & Galley, 1974). Ithas been noted that, in animals given a muscle-relaxant, electrical stimulation of ascendingspinal cord routes has no effect on vestibular efferents. Contrastingly, stimulation of thedescending motor axons from the reticular formation causes vestibular efferent activity shortlyafter (l5ms; Klinke & Galley, 1974). Both a feedback and a feed-forward mechanism have beenproposed for the efferent system. Plotnik et a!. (2005), using decerebrate chinchillas, put forwardone model that includes positive and negative feedback, and another that included a feed-forwardexcitation loop, monitored by descending inhibition. Currently, all models are highly theoreticaland the behaviour and purpose of the vestibular efferent system in conscious, intact animals (andhumans) is unknown.Reticulo- and vestibulo-spinal pathwaysAfferent information from the vestibulocochlear nerve is integrated with otherinformation in the vestibular nuclei, located in the medulla and pons in the brainstem. There arefour vestibular nuclei; superior, lateral, medial and inferior, on each side of the brainstem. As agroup, they accept input from 4 sources: the vestibular afferents (via the vestibulocochlearnerve), the vestibular cerebellum, the reticular formation and the contralateral vestibularafferents (via commissural fibres; Ruckenstein, 2004).The four vestibular nuclei are responsible for different aspects of vestibular control. Thelateral nucleus controls vestibulospinal reflexes throughout the body via the lateral16vestibulospinal tract (LVST). These reflexes are important for balance and ipsilateral extensortone. The medial and inferior nuclei together influence more proximal vestibulospinal reflexes,acting on the vestibulocollic reflex and general head and neck coordination through the medialvestibulospinal tract (MVST, also called the descending medial longitudinal fasciculusordescending MLF). Lastly, the superior and medial nuclei coordinate the vestibuloocularreflexthrough the ascending MLF (Blumenfeld, 2002).The vestibular nuclei also communicate with the brain. The nuclei output to the bulbarreticular formation and the vestibular cerebellum in particular, inhibiting flexors and extensorsthroughout all levels of the spinal cord via the reticulospinal tract (RST; Ruckenstein, 2004),which is known to be involved in automatic balance control (Blumenfeld, 2002). There are manymore vestibular connections within the brain, which are not the focus of this project andtherefore will not be reviewed here (see Carpenter, 1988, for further detail on these).Recent work by Cathers et al. (2005) used electrical vestibular stimulation to determinethat the otolith organs and SCCs can be considered as separate sensory systems, which causemuscular reflexes via independent reflex pathways. These pathways were previously postulatedto be the reticulospinal and lateral vestibulospinal tracts (Britton et al., 1993), based on anassessment of their vestibular connections and conduction speeds. A later study (Dakin et a!.,2007) suggested that the otolith signal traveled through the reticulospinal tract, and the SCCsignal traveled through the vestibulospinal pathways, based on frequency characteristics of thereflex response. However, work on cats (Uchino et al., 2005) has shown that 15-43% (perhapsmore) of the neurons in the vestibular nuclei responded to convergent input from more than onevestibular organ source. Of these, convergent inputs between otoliths and SCCs were common.Evidently, it is still unclear how vestibular inputs converge and are weighted to evokemuscular17reflexes. What is clear is that there are connections from the vestibular organs,through thevestibular nuclei (where other information may be integrated) to the muscles of the lowerlimbs.Vestibulomotor reflexMuscular reflexes in the limbs or trunk evoked by vestibular inputs are termedvestibulomotor or vestibulospinal reflexes, as they begin in the vestibular organsand arecommunicated to the muscles via the spinal cord. Most information on these reflexeshas beencollected using artificial vestibular stimulation (see “Galvanic vestibular stimulation”and“Stochastic vestibular stimulation” sections).Vestibulomotor reflex responses aren’t always present in all muscles. They arefoundonly in those that are posturally active, whether they are upper or lower limb(Britton et al.,1993). Muscles that are active but not involved in maintaining balance do not exhibitthis reflex,although motor neuron pool excitability may still be affected by the vestibularsystem (Kennedyet al., 2004).The vestibulomotor reflex response can be broken down into two parts, the shortlatency(SL) and medium latency (ML) waves. These waves are of opposing polarity, withthe MLresponse in the direction of resultant postural sway (Fitzpatrick & Day, 2004).In electricalstimulation studies, muscle facilitation is ipsilateral to the anode,and inhibition is ipsilateral tothe cathode. The SL wave has an onset of 55-65ms in the leg, andis smaller than the MLresponse. The ML response latency is about 110-l2Oms in the lower limb(earlier in theparaspinals at 61-75ms; Ali et al., 2003; Ardic et al., 2000; Britton et al., 1993).Balance is dependent on many sensory systems (as reviewed above), andit follows thatchanges in sensory conditions may change the size or Latency of the responses originatingin thevestibular system. Head position, for example, changes the meaning of vestibularinformation in18relation to the body. The polarity of the vestibulomotor reflex is therefore dependent on headposition (Britton et al., 1993; Dakin et al., 2007; Lund & Broberg, 1983), with resulting swayresponses directed along the interaural axis (Fitzpatrick & Day, 2004). Both SL and MLresponses increase in magnitude when vision is removed, external support is removed, thesurface is sway-referenced (or compliant), and when stance width decreases (Welgampola &Colebatch, 2001). These changes are all reflective of the relative importance of vestibularinformation under such sensory conditions. An increase in background muscle activity can alsoincrease reflex size (Lee Son et al., 2005).The ML response seems to be more susceptible to changes in sensory input than the SL,and more useful to maintenance of balance because of its direction (Fitzpatrick & Day, 2004). Itis not clear what purpose the SL response serves. Indeed, some researchers have gone so far as tocall it an evolutionary dead end (Britton et al., 1993) and many studies looking at vestibularorigin reflexes report only ML information. This view is not unanimous, however: one studyproposed that the origins of the SL and ML are separate, attributing the ML to the SCCs andconcluding that the SL was of utricular origin (Cathers et al., 2005). Neither viewpoint hasconclusive evidence to endorse it. The exact origin(s), pathway(s) and purpose(s) of thevestibulomotor reflexes are still uncertain. Results found using artificial vestibular stimulation,however, have contributed towards the further understanding of the behaviour of this reflex.Although this thesis does not use this technique, is worthwhile to present it in this review as it isuseful to understand the origin of the reflexes presented above.Galvanic vestibular stimulationGalvanic vestibular stimulation, or GVS, is a research method commonly used toinvestigate the vestibular system; vestibulomotor reflexes in particular. It consists of an19electrical current applied trans-cutaneously at the mastoid process(es), which transientlystimulates the vestibular system at the discharge site (where the nerve rootlets acceptneurotransmitters from the transducing hair cells; Goldberg et al., 1984). The firing rate in thenerve is increased or decreased, depending on whether the current is cathodal or anodal,respectively. The overall result in standing is sway towards the anode (Lund & Broberg, 1983).An important feature of GVS is that it is non-selective in its effect. That is, it will changethe firing rate of all vestibular afferents, no matter their origin. It has been found, however, thatirregular vestibular afferents are most sensitive to GVS, and its effect on these afferents faroutweighs its effect on regular and intermediate afferents (Goldberg, 2000). The cumulativeeffect of GVS on the vestibular organs is proposed in Fitzpatrick and Day’s 2004 review(Fitzpatrick & Day, 2004). It is believed that GVS is interpreted by the brain as a real vestibularsignal, and that its effects are compensatory to this (erroneous) signal. Sway, changes in walkingtrajectory and vestibulomotor reflexes can be all elicited using GVS in this way.Stochastic vestibular stimulationFollowing the success of pulsed GVS experiments, research groups began to use pseudo-random band-limited GVS (termed stochastic vestibular stimulation, or SVS) in order to evokesway and reflex responses in humans. Different research groups have used a form of frequencyanalysis (see the “Data Analysis” section for a more detailed look at this analysis) to investigatethe relationship between the SVS input and resultant EMG or CoP. In this way, it has been foundthat humans can act much like a “responder” to low-frequency bipolar SVS (0-2Hz) in themediolateral direction (CoP; Pavlik et al., 1999), and to low-frequency monopolar SVS in theanteroposterior direction (Scinicariello et al., 2002). SVS appears to have the same effect onpostural stability as profound bilateral vestibular loss (MacDougall et al., 2006).20Another group recently tested lower limb vestibulomotor reflexes elicited by SVS instanding. A 0-50Hz stimulus bandwidth was tested, and statistically significant coherence withlower limb EMG was found in the 0-20Hz range, maximally in the 5-7Hz and Ll-l6Hzbandwidths, for bilateral soleus and medial and lateral gastrocnemii (Dakin et al., 2007). Reflexpeaks and polarities corresponded to those elicited with GVS.It is apparent that electrical vestibular stimulation can elicit reflexes in posturally engagedmuscles, and that these reflexes originate in the vestibular organs and are frequency-dependent.It is unknown whether vestibular reflexes can be evoked by naturally-occurring vestibularstimuli; that is, accelerations at the head during quiet standing sway. The investigation of thisquestion was the basis for the second part of this thesis.21Research motivationPrior to this study, there was no complete description of the acceleration profile of thehead during quiet standing. Linear accelerations had been studied to an extent (Kitahara, 1965;Tsujikawa, 1966), but detailed analysis was lacking. There were no studies reporting angularhead acceleration behaviour during quiet standing at all. The complete description of linear andangular acceleration at the head during quiet stance under different visual and somatosensorysensory conditions was the first aim of this thesis.In his early experiments regarding body sway, Nashner (1971) calculated that thevestibular organs are sensitive enough to detect these small accelerations at the head during bodysway (which will be characterized in the first part of this thesis). Studies on vestibular-losspatients show that they cannot stand without somatosensory and visual inputs (Nashner et al.,1982). It is apparent that the vestibular system has some role in standing sway, but what role itmay play is unknown. Studying the relationship between accelerations at the head (what thevestibular system transduces) and forces at the feet may help to decipher what role the vestibularsystem plays in standing balance, and also uncover balance mechanisms that include the head.In 1994, Horak et al. eliminated vision and most somatosensory inputs and translated thehead of both healthy subjects and vestibular patients. They found direction-specific responses inthe lower legs that most likely stemmed from vestibular inputs, as these responses were notpresent in patients with adult-onset vestibular loss. This experiment showed that the vestibularorgans are coupled to the lower legs in absence of electrical stimulation. More recently, SVS hasbeen used to show that vestibulomotor reflexes can be elicited in the lower legs using stochasticelectrical vestibular stimuli. Dakin et al. used a correlation analysis to show a weak couplingbetween the artificial vestibular inputs and muscular responses in lower leg muscles in standing22(Dakin et al., 2007). It is reasonable to consider, then, that the vestibular system is able to beactively involved in the modulation of lower leg activity in quiet standing sway, and that it maybe activated by the head motions present during standing sway. The last part of this thesis wasdesigned to examine the relationship between lower leg muscle activity and angular headaccelerations in the direction of the semicircular canals during natural sway in quiet standingunder 4 sensory conditions.23AimsThe aims of this thesis study were as follows:A. To describe the angular and linear head acceleration behaviour in normal human subjectsduring quiet standing sway under 4 sensory conditions.Hypothesis1: Head accelerations characteristics will differ when measured under different sensoryconditions. R1\’IS will increase and median frequency will decrease as the standingcondition becomes more unstable.B. To determine the relationship between angular and linear head accelerations and forcesand moments at the feet, and to establish how this relationship changes with eye closure.Hypotheses1: Head accelerations and force plate forces will be coupled in-phase.2: Eye closure will most effect coherence below 1I-Iz.C. To determine the relationship between angular head accelerations and lower leg muscularactivity, and to establish how this relationship changes with changing sensory conditions.Hypotheses1: Angular head accelerations and EMG recordings from the lower legs will be coupled.2: This coupling will be strongest under the most vestibular-dependent sensory condition.24MethodsParticipantsEleven participants (7 female, 4 male) were recruited for the study. They were of goodhealth, with no past or current neurological conditions and no sensory or motor dysfunctions ofthe lower extremeties. Height ranged from 154 - 191 cm (mean 172 cm), and mass ranged from58-86 kg (mean 70.7kg). All subjects had intact vestibular systems and were between the ages of21 and 33 (mean 26.2 years old). This age range was chosen to exclude people over the age of 40because it is known that despite sensory conditions, focus or instructions, balance tends todegrade with age. Studies have shown that seniors (>65 years old) show an increase in COPexcursion over young subjects, and are more dependent on a hip strategy (normally reserved forlarge perturbations) than their younger controls (Amiridis et a!., 2003). Vestibular function alsotends to degrade as people age, leading to longer vestibulomotor reflex latencies and smallerreflex responses (Welgampola & Colebatch, 2002). These responses appear to remain intactbelow the age of 60 (Welgampola & Colebatch, 2002), however to err on the side of caution, nosubjects older than 40 years were recruited to participate in the study.All participants gave informed written consent, and the study was conducted inaccordance with the ethical guidelines established by the University of British Columbia.ApparatusAcceleration of the head was measured using a custom-made 9-accelerometer array. Arigid, lightweight, pyramidal array was attached to an adjustable plastic headband which wasfitted snugly to the participant’s head. All nine accelerometers were from Kistler InstrumentCorp. (Amherst, NY), model 833 0A3. These accelerometers were linear and had a working25range of +7- 3g and a sensitivity of 1 .2V/g (+1- 10%). They were recommended by Kistler forlow-acceleration, low-frequency applications, and they thus had a resolution of less than 1 .3jig,from 0-10Hz. The accelerometer output was DC and analog.Force plate data (forces in 3 directions and moments around 3 axes) was recorded usingan AMTI force plate (5571), in the same manner as the accelerometry data.Surface EMG was recorded from three different muscles, bilaterally. Soleus (SO), medialgastrocnemius (MG) and tibialis anterior (TA) were be recorded from the lower leg. Signalswere recorded using self-adhesive Ag/AgCl surface electrodes(SoftETMH59P: Kendall-LTP,Chicopee, MA, USA). They were placed over the specified muscle belly in landmarked areas,with an inter-electrode distance of—10 mm.The accelerometer and force plate data were recorded using a National Instruments PXI4495 DAQ, with 24-bit precision. Concurrent EMG data were collected from a NationalInstruments PXI-6289 DAQ, with 16-bit precision. Data were recorded simultaneously on thesame computer using a custom LabVIEW data acquisition program. A 1V square pulse from asignal generator was sent to both DAQ boards at the beginning of every trial to allow offlineelimination of computer lag between the boards. All data were recorded at 2kHz and saved in atext file to be analyzed offline; this analysis was completed using Matlab 7 software (MathworksInc., Natick, MA).The frame of the accelerometer array also supported a small plastic structure withreflectors mounted on it. This ‘head reference tool’ was part of the Polaris Vicra optical trackingsystem (NDI — Northern Digital Inc., Waterloo, ON). The tool, along with a reference wand andinfrared LEDs and detectors, allowed the accelerometry data to be transformed to head-spacecoordinates. All acceleration data was therefore able to be described in terms of Reid’s axes of26the head and, from there, could be further transformed to relate to the directional sensitivities ofthe SCCs.Experimental procedureTwo sensory conditions were manipulated in this experiment. They were the surfacecondition (hard or compliant) and visual condition (eyes open or eyes closed). Repeatability wasalso a concern, given the descriptive nature of the first aim. There were therefore 6 trials in total:2 control trials (both the first and last trial) and 4 experimental trials in between that wererandomized in order. Each trial was 4.5 uninterrupted minutes long (270 seconds). The trials aresummarized in Table 1. A brief familiarization session was provided before the first trial, whichacquainted the subjects to the task and all conditions (at approximately 30s per condition) beforethe accelerometer array was fitted. This allowed familiarization to occur without increasing theamount of time with the accelerometer array tightened (which tended to be uncomfortable afterlong periods). Seated breaks were offered after each trial, and a 10 minute break was observedafter the third trial (halfway through the testing session). At this point, the accelerometer arraywas removed to relieve head pressure, and subjects sat and rested. The accelerometer array wasreplaced and all points were redigitized before continuing to the last 3 trials.- . r1i •r’ I,, t .A. U r L1.J ILlIII LJ r1. r’ I p1i LII liti,,Control 1 Hard (force plate only) Eyes openExp 1 EOHS Hard (force plate only) Eyes openExp 2 ECHS Hard (force plate only) Eyes closedExp 3 EOSS Compliant foam Eyes openExp 4 ECSS Compliant foam Eyes closedControl 2 Hard (force plate only) Eyes openE- First trialRandom orderE— Last trialTable 1: List of experimental conditions27When subjects arrived for testing, they were fitted with electrodes for EMG collectionand stood on the appropriate surface for the short familiarization session. Thin metal washers(from 0 — 7 in number, as necessary) were added bilaterally between the hard frame of theaccelerometer array and the adjustable headband in order to fit each subject more securely andcomfortably. The accelerometer array, complete with head reference tool, was then tightenedaround the head, and the subject was seated with the Polaris Vicra system’s cameras in placebeside them. The reference wand was touched to 37 points to digitize them with respect to thehead reference tool. On the subject, these points were: nasion; the center of each externalauditory canal (to mark the location of the interaural axis); the points on the edge of the mostinferior part of each infraorbital rim; the inferiormost point of each mastoid process; skull vertex;external occipital protuberance; and glabella. On the accelerometer array, two points on oppositecorners of each accelerometer were digitized (18 points in all), as were 9 pre-marked points onthe aluminum frame (three on each orthogonal side; see Figure 1). In this way, Reid’s axes weredefined in terms of the position of the accelerometer array, and the center of mass of the headwas identified (National Aeronautics and Space Administration, 1978).Figure 1: Setup of the accelerometer array on a subject’s head. ‘11 e head reference tool is not seenin this photograph. Washers are visible between the hard and soft frames of the apparatus.28According to recent studies, the exact nature of the instructions given to the participantsaffects their sway patterns (Ishizaki et a!., 1991; Zok et al., 2008; Vuillerme & Nafati, 2007). Forexample, a person instructed to “stand relaxed” or “stand quietly” has a larger sway magnitudethan someone told to “stand as still as possible”. Therefore, subjects were instructed to standrelaxed on the appropriate surface (in bare feet or socks) with their arms by their sides and theirfeet as close together as possible without touching. During all “eyes open” trials, subjects wereasked to focus on a small, distant, stationary target at eye level.Since focus of attention can change sway patterns, subjects were not given any moreinstructions, nor were they exposed to undue external stimuli (i.e. music, talking, movingobjects, etc.). In a 2000 study, Shumway-Cook and Woollacott (2000) showed that during acognitive auditory tone-recognition task, young people (<45 years old) had no detriment topostural stability due to the additional attention demands, no matter the sensory condition.However, another study in 2003 showed (in young, healthy subjects) decreases in mediolateralsway magnitude during a cognitive task (a multi-step arithmetic problem), and increases in thesame value during a non-balance related motor task (a finger pinch at 10% MVC; Weeks et al.,2003). This study showed that directing a person’s attention internally (as in the finger pinch)may increase sway, while an external focus (the math problem, which was physically unrelatedto the person) decreases sway in young healthy individuals (see also McNevin & Wuif, 2002;Vuillerme & Nafati, 2007). In this study, attention was not purposely directed.Data analysisProcessing of the data involved rectifying all EMG channels and calculating linear andangular head accelerations for the axes of interest from the nine-channel accelerometry data. Thespatial data from the Polaris Vicra system was used to create a transformation matrix from an29array-space reference frame to one centered about Reid’s axes (modified as described in theliterature review). The origin of Reid’s axes was defined as the center of mass of the head. Thisis approximated to be in the midsagittal plane, rostral of the interaural axis by 17% of thedistance measured between the vertex and the interaural axis (National Aeronautics and SpaceAdministration, 1978; as in Blouin et al., 2007).Angular accelerations of the head were computed using the following equation set:A1 — A0A3 — A0= 2d1— 2d3A3 — A0 A2 — A0=2d3— 2d2A2 — A0A1 — A0= 2d2— 2d1Equation 1: Angular acceleration equation for the head, where: a is angular acceleration aboutaxis i, A1 is linear acceleration from accelerometer i, and d1 is the distance between accelerometers,from Blouin et al., 2007.Once angular accelerations relative to the accelerometer array axes were calculated, theywere transformed into accelerations about Reid’s axes, and also to accelerations about the axis ofeach SCC (as measured in Della Santina et al., 2005). Angular accelerations were high-passfiltered at 0.2Hz in an attempt to remove drift due to temperature and tilt, which appeared to beconcentrated below this limit. These data were also low-pass filtered at 100Hz. Linearacceleration data were transformed to the directions of Reid’s axes, and the DC offset wasremoved for all calculations except absolute maximum linear acceleration (as the effect ofgravity that it implies cannot be identified using this experimental setup).30These accelerometry data were used to complete the first aim of the thesis: describing theacceleration profile for the head. For both linear and angular accelerations in all 3 directions (inrelation to Reid’s axes) absolute maximum value, RMS, median frequency and the 95%confidence interval frequency were calculated. Of these, RMS and median frequency werecompared between the control trials (see Table 1: Control 1, Control 2 and Exp 1), and alsobetween sensory conditions 1 through 4 (see Table 1: Exp 1-4) as described in the Statisticssection.Following this description, a coherence analysis was completed between angular headaccelerations in the Reid’s planes and forces and moments collected from the force plate, inorder to uncover a possible linear relationship between them. Linear head accelerations (ML andAP) were run against forces (in terms of what the body, not the plate, experiences) in equivalentaxes. Angular head accelerations about all 3 Reid’s axes were also run against angular force platedata (the moments). Finally, linear head accelerations were run against force plate moments inthe direction that they were likely to affect (e.g. linear head acceleration in x may cause a forceplate moment about y). It is important to note that these coherence analyses were not valid forexperimental conditions 3 and 4 (those on the compliant surface), because it is unknown how theintroduction of the soft foam affects the transmission of forces from the subject’s feet to the forceplate. Therefore, force plate data from conditions 3 and 4 were not analyzed. Also, 3 of 11subjects had unusable force plate data. Their results are not included in any analyses using forceplate data; therefore, the force plate coherence analysis is completed using 8 subjects.A coherence analysis was also completed between angular head accelerations in theplanes of the SCC’s and all 6 muscle EMG recordings under each of the 4 sensory conditions,using data from all 11 subjects.31Coherence between the concurrent data sets in each analysis was computed over subjectsusing Equation 2 (below) on concatenated data sets of equal length from each subject. Coherenceis an indicator of the linear relationship between two signals across frequencies. It is unit-lessand bound between 0 and 1, 1 denoting a perfect linear relationship and 0 denoting independence(Halliday & Rosenberg, 1999; Rosenberg et al., 1989). The result can be interpreted as thepercentage of the signal’s variance that can be accounted for by the influence of the other signal.Coherence between the data sets was calculated using a publicly available Matlab script(available for download at http://www.neurospec.org/welcome.html), which is based on amethodology described by Rosenberg et al. (Rosenberg et al., 1989; used previously in Dakin etal., 2007 and Roskell et al., 2007). Final coherence plots were all subjected to a moving averagefilter using a Hanning window with weights of 0.25, 0.5, 0.25. This created a smoother plot andsimplified the identification of the major areas of coherence, without requiring modification ofthe 95% confidence intervals (Farmer et al., 1993).2— IfAB(A)12IRAB (A)I— fAA(A)fBB(A)Equation 2: Coherence between signals A and B is calculated by dividing the cross-spectra’smagnitude squared, by the product of the two autospectra.Time-cumulant density (TCD) functions were also calculated for the concatenated datasets to provide a time domain representation of the relationship between the signals. TCDfunctions are inverse Fourier transforms of coherence and although they cannot be used todirectly measure the amplitude of the relationship in millivolts, the timing of the function peaksand troughs (maximums and minimums, in this case) provides an estimate of the phase lagbetween the signals. The polarity of the response can also be interpreted from the plot. A positive32TCD indicates in-phase forces/accelerations between the force plate and head, or facilitation ofthe muscle in relation to head accelerations (in the EMG coherence analysis). A negative TCD ofcourse implies the opposite. TCD plots were not subjected to a filter.StatisticsStatistical comparisons on the acceleration results were made on two sets of data. Withinthese sets only RMS and median frequency were statistically tested. The first set, made of threecontrol trials, were compared using a 2-way mixed intraclass correlation (yielding an intraclasscorrelation coefficient or ICC) to test the reliability of the head acceleration measures. Themeasures were deemed repeatable if the ICC > 0.5 (Weir, 2005). The second set, of the foursensory condition trials, were tested for statistically significant differences using a 2-wayrepeated measures ANOVA. This ANOVA tested for main effects of surface and vision and foran interaction between them. If an interaction was present, a post-hoc Tukey test was performed.All tests were completed to a p value of 0.004, which is equivalent to ap value of 0.05 with aBonferonni correction factor for multiple comparisons of 12. The intraclass correlations werecalculated using SPSS (version 14.0, SPSS Inc., Chicago, IL); the ANOVAs were done usingStatistica (version 6.1, Statsoft, Tulsa, OK).Coherence analyses between angular head accelerations and force plate forces werecompleted on concatenated data (8 subjects total), to a frequency resolution of 0.015 Hz, usingsegments of 65.536s length. Analyses on angular accelerations used 33 of these segments, whileanalyses on linear accelerations used only 30 (due to the necessary shorter length of the linearaccelerations data set; it was shortened in transformation into Reid’s axes). Coherence betweenhead accelerations and EMG were also done on concatenated data (11 subjects total), to afrequency resolution of— 0.06 Hz, using 169 segments each 16.384s in length.33Any coherence was deemed significant at a particular frequency when it surpasseda 95%confidence interval, which is calculated based on the number of disjoint sections(Halliday et al.,1995; Halliday & Rosenberg, 1999). Differences in coherence between conditionswerecompared using a Difference of Coherence (DOC) method from Amjad et al. (Amjadet al.,1997). Ninety-five percent confidence intervals for the DOCtest were determined by a chisquared distribution with k-i degrees of freedom (k being the number ofconditions beingcompared; p=O.O5).34ResultsLinear and angular head accelerations about Reid’s axes were analyzed for absolutemaximum value, RMS, median frequency and 95% confidence interval frequency (95% CIF).Absolute maximum and 95% CIF are reported in the included tables but were not statisticallytested.Control trialsThe three control trials revealed repeatability in angular RMS and median frequency(ICC 0.867 in all directions for RMS; ICC 0.685 in all directions for median frequency). Theaverage RIvIS in roll, pitch and yaw was 2.02 ± 1.38 rad/s2,2.38 ± 1.08 rad/s2 and 2.39 ± 1.71rad/s2 (mean ± SD), revealing a high standard deviation in every direction. Average medianfrequency for the 3 control trials in roll, pitch and yaw was 8.5 ± 2.36 Hz, 3.94 ± 0.80 Hz and4.66 ± 0.62 Hz, respectively.Linear acceleration results were not repeatable in some instances. RIVIS was repeatable inAP and IS directions (ICC> 0.634) but was not in ML (ICC = 0.377). Average RMS for AP was0.21 ± 0.09 m/s2,and for IS was 0.06 ± 0.04 rn/s2.Median frequency was repeatable in AP(ICC=0.515; average value 0.02 ± 0.03 Hz,), but not in I\’IL or IS (ICC 0.382).Complete tables of control trial data can be found in Appendix 1.Sensory condition trialsFour experimental trials tested head acceleration characteristics under differentmanipulations of two factors: vision and surface. To recall, these trials were eyes-open hardsurface (EOHS, Exp 1), eyes-closed hard-surface (ECHS, Exp2), eyes-open soft-surface (EOSS,Exp 3), and eyes-closed soft-surface (ECSS, Exp 4). The results from these conditions are35summarized in Table 2 (angular) and Table 3 (linear). Convention follows our modified Reid’saxes: positive roll is right ear down, positive pitch is nose up, and positive yaw is nose right.Angular accelerationThe angular acceleration RMS ANOVA showed an interaction effect between vision andsurface factors for roll (F(1,1O)14.781), pitch (F(1,1O)=23.919) and yaw (F(1,1O)=19.612). Apost-hoc Tukey test revealed that in all axes, the difference lay in ECSS RMS, it beingsignificantly higher than RMS in all other conditions (roll all p<O.000393; pitch all pO.OOO2l3;yaw all p<O.000276; see Figure 2).The ANOVA on median frequency of angular head accelerations showed main effects ofsurface in roll (F(1,1O)24.135;p=O.00061)and yaw (F(1,1O)26.373; p=O.00044), wheremedian frequency under HS conditions was significantly higher than SS. No main effects inmedian frequency were observed about the pitch axis. Results can be seen in Figure 3.36Exp 1EOHSExp 2ECHSExp 3EOSS_______ROLL (x)__________________Absolute Max 1.88+1-1.09 2.14+1-0.92 2.29+/-1.04 3.57+1-2.11 rad/s2Overall RMS 0.27 +1- 0.06 0.29 +1- 0.06 0.31 +1- 0.05 0.39 +1- 0.07 rad/s2Median Freq 8.45 +1- 2.50 8.10 +1- 1.65 7.20 +1- 2.16. 5.14 +1- 1.19 Hz95% CIF 21.35+1-2.75 20.50+1-2.60 20.06+1-1.86 18.79+1-2.53 HzPITCH(y)Absolute Max 2.42+1-0.90 2.39+1-0.91 2.63+1-0.57 4.05+1-1.24 rad/s2Overall RMS 0.29 +1- 0.05 0.331-/- 0.05 0.36+1-0.06 0.51+1-0.08rad/s2Median Freq 3.95 -i-/- 0.80 3.69+1-0.57 3.69+1-0.61 3.69+1-0.48 Hz95% CIF 17.66 +1- 3.31 16.73+1-3.42 14.09÷1-2.14 11.88÷1-1.19 HzAbsolute Max 2.21÷1-1.61 2.33+1-1.71 2.51+1-1.47 3.81+1-2.11 rad/s2Overall RMS0.29+1-0.06 0.31÷1-0.06 0.37+1-0.07 0.49+1-0.09rad/s2Median Freq 4.62 +1- 0.50 4.41 +1- 0.47 3.96 +1-0.42 3.60+1-0.45 Hz95% CIF 15.56+1-2.39 15.38+1-2.22 14.11+1-2.25 12.49+1-2.19 HzTable 2: Angular acceleration characteristics in roll, pitch and yaw under four sensory conditions.Mean +1- standard deviation.37Exp 4ECSSYAW(z)Angular Acceleration: RMS ± SD*_I IRoll Pitch YawFigure 2: Angular acceleration RMS under four sensory conditions in three directions (mean ±SD).Presented as same-shaped points in order, these are: EOHS, ECHS, EOSS and ECSS. Thestatistical analysis showed an interaction between vision and surface factors; a post-hoc Tukey testrevealed that RMS under ECSS was significantly larger than RMS in all other conditions(p<O.000393, denoted with an asterisk).Angular Acceleration: Median Frequency ± SDN>.Ua)a)LI.CCoa)r.1IflCoC0.4-COIa)a)UU2COzCC,) under ECSS conditions is significantly largerthan RMS under all other conditions, in all directions.121086420B Hard SurfaceSoft SurfaceB Hard SurfaceSoft SurfaceEyes Open Eyes ClosedFigure 3: Angular acceleration median frequency in three directions (mean ±SD). The statisticalanalysis showed main effects of surface condition in roll and yaw (pO.OOO6l, denoted with anasterisk), but no main effects in pitch.38Linear accelerationThere were no significant main effects in the linear acceleration data, either in RMS ormedian frequency (all p>O.OO756). Table 3, below, shows the values of the linear accelerationmeasures under all sensory conditions.Absolute MaxOverall RMSMedian Freq95% CIFAbsolute MaxOverall RMSMedian Freq95%CIF_____________ ________AP (x)2.57÷1-1.07 2.91+1-1.16 2.50÷1-1.12 2.60+1-1.150.24+1-0.09 0.20+1-0.08 0.21+1-0.06 0.27+1-0.090.01+1-0.02 0.04+1-0.02 0.03+1-0.02 0.05+1-0.101.17+1-1.58 1.98+1-1.63 1.92+1-1.57 1.73+1-0.76ML(y)0.73÷1-0.25 0.85 -i-I- 0.26 0.93+1-0.38 1.34+1-0.720.11÷1-0.05 0.12+1-0.04 0.13+1-0.03 0.19+1-0.080.05+1-0.06 0.06+/-0.05 0.07+1-0.04 0.15+1-0.129.93+1-5.74 9.29÷1-5.46 7.67+1-5.71 4.52+1-4.11rn/s2rn/s2HzHzrn/s2rn/s2HzHzAbsolute Max 9.79 -i-I- 0.18 9.76+/-0.20 9.85÷1-0.19 9.97+1-0.23 rn/s2Overall RMS 0.06 -t-/- 0.04 0.06+1-0.04 0.05÷1-0.03 0.07+1-0.04 rn/s2Median Freq 0.74+1-2.14 0.27÷1-0.40 0.68÷1-1.44 1.40+1-1.69 Hz95% CIF 17.68 ÷1- 11.26 15.93 +1- 9.11 17.22 +1- 9.65 15.08+1-8.47 HzTable 3: Linear acceleration characteristics in anteroposterior (AP), mediolateral (ML) andinferiosuperior (IS) direction under four sensory conditions. Mean ± standard deviation.Canal-transformed accelerationsAngular and linear head accelerations about Reid’s axes were transformed to thedirections of the right SCCs using transformation matrices from Della Santina et al. (2005). Thevalues are reported in Table 4 on page 40 for discussion. The results are statistically comparedbetween canals in Appendix 2.Exp 1EOHSExp 2ECHSExp 3EOSSExp 4ECSSIS(z)39Absolute MaxOverall RMSMedian Freq95% CIFAbsolute MaxOverall RMSMedian Freq95%CIFExp 1 Exp 2 Exp 3 Exp 4EOHS ECHS EOSS ECSSZ1TF1IJ2.22÷1-0.85 2.29+1-0.65 2.49÷1-0.81 4.00+1-1.470.28 -f-/- 0.05 0.31+1-0.05 0.35÷1-0.05 0.50 -i-I- 0.074.94+1-0.90 4.55 -f-f- 0.60 4.46+1-0.82 4.00+1-0.5020.04+1-2.55 19.48-‘-I-3.11 17.64÷1-2.77 14.85+1-2.102.20+1-1.51 2.44÷1-1.79 2.60÷1-1.29 3.81+1-1.870.30÷1-0.06 0.32 -i-/- 0.06 0.37+1-0.07 0.49÷1-0.084.95÷1-0.69 4.93÷1-1.14 4.04 -i-/- 0.52 3.55+1-0.4718.10+1-2.51 17.76÷1-2.34 16.42÷1-1.61 15.01+1-1.681.76÷1-0.79 2.03+1-0.98 2.34+1-0.830.27+1-0.06 0.29+1-0.05 0.32+1-0.055.46÷1-0.97 5.06÷1-0.85 4.86 -i-/- 1.0519.05+1-2.17 18.24÷1-2.88 16.78 -t-/- 1.743.48+1-1.920.43+1-0.084.15+1-0.6814.60+1-1.69Right Posterior Canal (z)Absolute MaxOverall RMSMedian Freq95% CIFTable 4: Angular head acceleration characteristics in the directions of the canals under foursensory conditions. Mean +1- standard deviation. Statistical differences are not marked.CoherenceForce plateCoherence was first calculated to uncover possible linear relationships between certaincombinations of head acceleration and force plate forces. The results can be seen in Figure 4.Linear head accelerations vs. linear force plate forces in equivalent axes were first examined, inanteroposterior and mediolateral directions. Coherence between linear head acceleration andlinear force plate forces in the anteroposterior direction (“AP relationship”) reached significanceunder 7Hz, with three distinct peaks: one below 1Hz (maximum coherence 0.49), one at 1-3Hz(reaching a coherence of O.36) and a larger one at 4-7Hz (reaching a maximum coherence ofRight Horizontal Canal(y)rad/s2rad/s2HzHzrad/s2rad/s2HzHzrad/s2rad/s2HzHz400.54). A Difference of Coherence (DOC) test revealed that coherence increased significantlywhen vision was removed, both under 1 Hz and in the peak around 5Hz. In the mediolateraldirection (“ML relationship”), coherence peaked in the very low-frequency range (under 1 Hz;coherence 0.59) and again at 3.5 Hz. A DOC test showed that this 3.5 Hz peak was significantlylarger in the EO condition (coherence of 0.46 vs. 0.35 in EC), and again showed significantdifferences in coherence under 1Hz (EC> EO).Coherence between angular head accelerations and plate moments was calculated usingaxes in the same direction. In roll, small peaks (coherence -0. 12) around 3, 5 and 7Hz areobserved under both conditions. The pitch direction also shows low coherence in similar regions(coherence 0.15-0.20). The DOC test revealed that pitch coherence at 3 and 7Hz was higherwhen the eyes were open. Both pitch and roll also had significant coherence under 1Hz, whichincreased with eye closure.Coherence in the yaw direction was seen in three distinct areas: under 2Hz (peak at acoherence of 0.69), at 3.5Hz (peak coherence 0.35), and 5Hz (peaks coherence from 0.2-0.3).Small groups of peaks (coherence <0.2) were also seen at ‘—‘7Hz and ‘--‘12Hz, with significancefound up to 20Hz (the upper frequency limit tested). A DOC test revealed differences under 2 Hzand that eye closure caused a lower 5Hz peak.The last set of data analyzed for force plate coherence were linear head accelerations vs.the force plate moments that they would cause, given an inverted pendulum model about theankles. Coherence between linear head accelerations and angular force plate data reachedsignificance in certain bands in both directions. The linear acceleration of the head in theanteroposterior direction (x) was related to the force plate moments about the pitch axis(y)differently according to visual condition. Both visual conditions showed some significant41coherence from 0-20Hz, but areas of maximum coherence were concentrated below 2Hz andbetween 4-6Hz. The DOC test showed that eye closure increased coherence under 2FIz(coherence —0.50 in EC vs. —0.25 in EO), and in narrow-bandwidths at 4, 6 and 9Hz.Differences between conditions also appeared in coherence between linear headaccelerations in the mediolateral direction (y), and angular plate forces about the roll axis (x).The EO condition was coherent from 0-20Hz, with peaks under 1.5Hz (coherence —0.57), and ataround 6, 10, and 15Hz (all coherence —0.25). In the EC condition, coherence reached 0.60 under1.5Hz, and a distinct low-coherence peak was found between 5-6Hz. A DOC test showed oncemore that eye closure increased coherence below 1Hz, with other small differences showing thatcoherence decreases with eye closure in the higher frequencies.Time-cumulant density (TCD) plots revealed the timing and polarity of the relationships.When considering the implications of polarity it should be noted that, mechanically speaking, theforces at the feet must be proportional to the acceleration of the center of mass in the samedirection. Since signal magnitude is not of importance to the coherence analysis (except whenconsidering the signal-to-noise ratio), one can consider the coherence between foot forces(/moments) and head accelerations to also be representative of the coherence between center ofmass linear (/angular) acceleration and head accelerations. TCD polarity signals the phaserelationship: positive indicates in-phase (e.g. forward/forward), and negative indicates out-ofphase (e.g. forward/backward).TCD plots from pooled data revealed that AP plate forces (forces on the feet) appeared tolag AP head accelerations by approximately 180 ms in the EO condition, and 145ms in the ECcondition. In the ML direction, however, foot force preceded the head accelerations by 21 Oms inthe EO condition and by 11 Oms in the EC condition. In both AP and ML directions the TCD was42positive, meaning that the forces on the feet are in phase with head accelerations(forward/forward, etc.).TCD plots also showed that force plate moments in roll preceded angular headaccelerations about the same axis by about 2lOms EO, and l3Oms EC. In pitch and in yaw, theforce plate moments occurred after the angular head accelerations about the same axis. In pitchthis lag was 100 — 110 ms (EC/EO, respectively). In yaw, the lag decreased to about 45ms inboth conditions. The relationships were negative in polarity, indicating opposite phase. In thiscase, however, it is the moments on the force plate that we compared, not those on the foot.AP head accelerations preceded pitch plate moments by 65ms in the EO condition, andby 5ms in the EC condition. The signals were in-phase. In contrast, ML head accelerationslagged angular plate moments in roll by about 80 — 90 ms (EC/EO, respectively), and the signalswere out of phase. TCD plots can be seen in Figure 4.Interestingly, the plane of sway seemed to dictate whether the relationship betweenforces/moments and linear/angular head accelerations was a leading or lagging one. In thesagittal plane, which includes AP forces and pitch moments, the head accelerations lead forceplate forces. In contrast, in the frontal plane including ML forces and roll moments, headaccelerations lagged force plate forces, in similar time frames.The differences in EO/EC are not statistically significant according to dependent t-testsconducted on data from all subjects (8 subjects; allp> 0.05). Statistical tests could not be run ontimings from pooled data between conditions as there were only two values to compare.43Head Accelerations vs. Force Plate Data: Coherence, DOC and TCDCoherence DOC Time-Cumulant DensityLinear Accel Rxvs.Linear Plate FxEOHS ECHSQo4o_Th:0tFigure 4: Coherence and TCD plots for Reid head accelerations vs. force plate forces. Red linesdenote 95% confidence intervals for coherence (blue EO, green EC) and DOC. Black crossesrepresent zero in the TCD plots; blue lines represent parts of the plot that do not reach significance.Linear Accel RyvsLinear Plate FyAngular Accel Rxus,Angular Plate MxAngular Accel Ryvs.Angular Plate MyAngular Accel RzVS.Angular Plate MzLinear Accel RxVs.Angular Plate MyLinear Accel Ryus,Angular Plate Mx10 200 5 -5 5———Freq (Hz)————— Time (s)44EMGCoherence was calculated between angular head accelerations in the directions of allSCCs, and EMG from soleus (SO), medial gastrocnemius (MG) and tibialis anterior (TA)bilaterally. Select results are presented here, but caution is advised. The paper from which thecanal transformation matrix was taken (Della Santina et al., 2005) explicitly warned againstusing the results to calculate prime directionality of the canals from anatomical landmarks, asthey are poorly aligned. The assumption that these prime directionalities are reliable fromlandmarking is inherent in the interpretation of the coherence analysis. Results are therefore notlikely to be robust. They are selectively presented in short-form (soleus only) in Figure 5, andcan be found in their entirety in Appendix 3.Plots of coherence between bilateral soleus and the 3 SCCs reveal a few interestingrelationships. There are significant peaks in coherence between the horizontal canal accelerationand bilateral soleus under 6Hz (up to a coherence of 0.16 in ECSS). A DOC test revealed thatcoherence increases significantly in parts of this region from EOHS through ECHS and EOSS, toECSS.In the anterior and posterior (or together, vertical) canals, coherence differed by muscleside. A 4-way DOC test revealed no difference between conditions in right soleus/anterior canalcoherence, or between conditions in left soleus/ posterior canal coherence. Differences were seenin left soleus/anterior canal coherence and right soleus/posterior canal coherence, however,between all conditions. This may be due to signal strength. Forward-right accelerations, whichthe right anterior canal would pick up, would require excitation in right soleus to counteract inorder to maintain stability. Forward-left accelerations would require excitation in left soleus tomaintain stability and would be picked up by the left anterior canal, which is approximately45equivalent in direction to the right posterior canal. Active contraction would increase the signalto noise ratio, making a significantly coherent result more likely. This side will be termed the“leaning-affected side.”In both vertical canals, the leaning affected side was highest in coherence under theECSS condition but differed in areas between all conditions. In ECSS, significant coherence wasfound in two main peaks, from 0-2Hz and 3.5-7.5Hz, with coherence reaching only 0.17 atmaximum. A less well-defined peak also is visible around 17-20Hz. The lower-frequency peakseems to correspond with SVS/EMG coherence results from Dakin et al. (2007), who reportedpeaks from 5-7Hz and 11-16Hz. Dakin’ s higher-frequency coherence band was not observed.46a)0Coherence: Head acce’eration in SCC directions vs. bilateral soleusPooled Coherence Estimate DOC 4-way)0.2EOHS ECHS EOSS ECSSPast0.2Hor0.2Antiaiqaijiiiii kLri0.2Post0.2Hor0 20 40za)0Figure 5: Head accelerations in the direction of the SCCs (anterior, posterior and horizontal canalsas marked) vs. EMG from bilateral soleus. Horizontal lines denote 95% confidence intervals,sensory conditions are as labeled. The rightmost plot is the 4-way DOC test.TCD plots revealed a lead of muscles over head accelerations in all canals (see Appendix3). The relationship in soleus for all canals is shown in Figure 6. The muscle lead is seen0 20 40—Freq (Hz)0 20 4047bilaterally, with an opposite polarity in opposing legs. In soleus, for the EOHS condition in theanterior canal, this lead is approximately 1 35ms for the first peak (the one furthest from timezero). The second peak (closer to time zero) is of opposite polarity and occurs before headaccelerations by around 4Oms. In the horizontal and posterior canals, EMG leads headaccelerations by around 115-125ms, as calculated from the pooled TCD plots. These lead timeschange in the ECSS condition, as evidenced in Figure 6. Under these conditions, the first peak inright soleus EMG leads head accelerations in the anterior canal by 360ms, in the posterior canalby 200ms, and in the horizontal canal by 330ms.Time Cumulant Density: Head Acceleration & EMGEOHS ECHS EOSS ECSSzCHorz____fTime (s)Figure 6: TCD plots between anterior, posterior and horizontal semi-circular canal accelerationsand soleus EMG under four sensory conditions. Black lines are right muscles, grey are left. Crossesdenote zero. Blue parts of the plots do not cross the boundaries of significance.Ant-1 048DiscussionRepeatabilityThe nature of this study is exploratory in that angular head accelerations have not beenreported in literature to date. If this data is to be compared to other populations (patients, forexample) in the future, then repeatability in healthy subjects is certainly of interest. According tothis study, angular head acceleration RIVIS and median frequency were repeatable in all axesunder the full sensory condition (EOHS; about Reid’s axes). Linear head acceleration measureswere repeatable in the anteroposterior direction but were not mediolaterally. This may bebecause movement in the anteroposterior direction has the advantage of a larger base of support,inevitably increasing RMS and median frequency standard deviations and therefore lowering thechances of finding a significant difference between trials. Having said this, it is not clear whetheror not this non-repeatability is actually true for linear head acceleration measures, given somelimitations in linear acceleration measurements (discussed in “Limitations”).Repeatability was not tested under differing sensory conditions, and therefore it cannot becommented on past EOHS.Comparison of data to previous accelerometry workThere have been no prior studies against which to compare angular head accelerationmeasurements, but linear head acceleration reports are available. Previous studies report meanmediolateral head accelerations in the EOHS condition from 0.Ollm/s2(Winter et al., 1998) to0.2m/s2 (Tsujikawa, 1966). The values from this study more closely resemble Tsujikawa’sresults, with linear head acceleration RMS around 0.11—0.1 9m/s2.This is possibly due toTsujikawa’s direct measurement technique. His group used accelerometers, while Winter et al.49used displacement data to calculate accelerations. Tsujikawa also measured brief periods ofmaximum acceleration in the mediolateral direction (with the eyes closed in a tandem stance) ofapproximately 2.45m/s2.In the roughly equivalent stance in this study (ECSS), maximumaccelerations reached 1 .34m!s2 in ML and nearly 3m/s2 in AP. Of course, linear accelerometerplacement may contribute to this difference, and these directional discrepancies might bepartially owed to Tsuj ikawa’ s methods, specifically the use of a tandem stance and one linearaccelerometer atop a helmet. The conclusion is that linear acceleration measurements seem toreplicate those previously found.Comparison of acceleration data to reported vestibular thresholdsThe most important message obtained from analyzing the sensory condition trials wasthat both absolute maximum and RMS angular head acceleration measured in this studyexceeded reported thresholds for the vestibular SCCs. Subjective thresholds are reported from0.06-3 deg/s2,depending on the study, while thresholds established by oculogyral illusionmeasurement found smaller values, around 0.04-0.28 deg/s2 (reviewed in Nashner, 1971).Nashner (1971) argued that these values varied because of differing response modalities, andendeavoured to discover the SCC threshold for the postural response modality. He calculated thisthreshold to be 0.O5deg/s2,using an accelerated platform with a nulled ankle angle and no vision.Evidently, this setup is not equivalent to quiet standing, but does yield an interestingly lowvestibular threshold ‘in a postural response modality’. In any case, the smallest angular headacceleration RMS found in this study (that in the EOHS condition in the right posterior canaldirection) was 0.27 rad/s2;approximately 14.9 deg/s2.This is much larger than any reportedvestibular threshold in any of the response modalities, and values become increasingly larger asvision and surface factors are manipulated. It is apparent that the vestibular system is50physiologically able to detect most of the angular acceleration behaviour at the head during quietstanding.Angular head acceleration median frequency (in SCC directions) varied from around 3.5-5.5 Hz, over all conditions and directions. Head acceleration frequency measures to compare toare not reported in the literature to date. Displacement frequency values are more common butwere not accurately calculable from this study’s data. They would, however, certainly not behigher in frequency, so it is likely that these frequency values still fall into the range at whichHighstein (2005) and Lysakowski & Goldberg (2004) argue that the SCCs are velocitytransducers (0.025Hz — 30Hz). Ross (2003), however, has found collateral inputs between haircells in the SCCs that likely lead to on-line differentiation. Acceleration data may still betransduced in the SCCs in quiet standing.Linearly, the smallest head acceleration RMS occurred in the inferiosuperior direction,and was around the threshold for the otolith organs as reported by Meiry (1966). RMS in AP andML directions greatly exceeded this threshold (0.20 and 0.11 m/s2,respectively, vs. Meiry’sthreshold of 0.059m/s2),therefore it is likely that the otolith organs are able to detect linear headaccelerations, in the transverse plane at least, that occur during quiet standing. It is thought thatthe otolith organs are most sensitive at a low frequency, and therefore act as a static and low-frequency vertical reference during quiet standing (Nashner, 1971). This argument may besupported by the results of this thesis, given the low median frequency found in the linear headacceleration data, which likely reflects a gravitational position effect.The inverted pendulumIt is recalled from the literature review that the postural strategies used in standingbalance are the ankle and hip strategy. The ankle strategy results in inverted-pendulum-type51sway and is normally used under full sensory conditions (Winter, 1995). Subjects resort to thehip strategy, which involves counterrotation of the head and hips, under unstable or shortenedbase of support conditions. The use of these strategies during the trials can be alluded to from theforce plate/head acceleration TCD data. The positive polarity of both the AP and MLrelationships reveals that the CoM and the head accelerate in-phase linearly, which seems topoint to the use of an inverted pendulum ankle-strategy. In addition, the plate moment/angularhead acceleration relationships also suggest an inverted pendulum. Although the polarities werenegative, they were calculated using forces with respect to the force plate. When using themoments the foot (and consequentially the CoM) experiences instead, these polarities would bepositive, pointing to an in-phase angular relationship as well. These findings are supported by astudy by Winter et al. (2003), in which the applicability of the inverted pendulum model wasinvestigated in each plane. The group found that the height and horizontal displacement RMS ofbody markers were correlated with a linear regression through the points at an R2 of 0.966 in APand 0.944 in ML (Winter et al., 2003). They concluded that the inverted pendulum model wasvalid in both planes, as we do.The hip strategy involves counterrotation of the head and hip (Horak & Macpherson,1996). If it was dominant, the polarity expected from the force plate relationships would beopposite to that found in the EOHS and ECHS conditions. Unfortunately force-platerelationships could not be investigated under soft surface conditions (see Limitations), so theexpected negative polarity could not be established. However, a decrease in angular headacceleration median frequency on soft surfaces suggests a hip strategy and it was observationallyapparent in the subjects during the soft surface conditions.52Although the relationships in the force plate TCD data suggested inverted pendulumsway on the hard surface in AP and ML, the analysis revealed that the timing differed betweensagittal and frontal planes. Head angular accelerations lead moments on the feet in pitch (sagittalplane) by around lOOms, but lagged them in roll (the frontal plane) by 130-2lOms. This findingpoints to differing balance mechanisms in the AP and ML directions, which was originallysuggested by Winter et a!. (1996) in a CoP study. They alleged that the balance mechanismswere independent in the different planes (owing to biomechanical differences). This thesis showsthat the difference between these independent mechanisms lies in the timing of control in eachplane. A “top-down” mechanism is indicated in the sagittal plane, where the head appears toaccelerate just prior to moments appearing at the feet. A “bottom-up” mechanism describes thefrontal plane, where moments at the feet precede head accelerations in roll. Further, this impliespassive head control in the frontal plane, as the head accelerates after moments appear at the feet.Sagittal head control, being opposite in timing, is suggested to be active in a quietly standingperson.These conclusions are, of course, based on the relative behaviour at only two points andare valid only on a hard surface. Further study on the timing of the accelerations of more bodysegments would aid in strengthening or refuting this idea.Purposeful swayThere are many ways in which quiet standing sway could be composed of both reactiveand proactive contributions. Reactively, external (gravity, environmental interactions) andinternal (respiration, heartbeat, articulation) perturbations may ‘knock’ the body, moving thecenter of gravity and requiring correctional sway back to the middle of the base of support.Indeed, contributions from heartbeat have been seen in linear head acceleration plots (Kitahara,531965; Tsujikawa, 1966). Conversely, from a proactive viewpoint sway can be considered not as aconsequence of constant corrections, but instead as a useful tool for creating a robust sensoryenvironment. If possible, a completely static standing posture would seem ideal in regards toenergy conservation and neural simplicity, but such a posture would deprive the body of a streamof constantly changing afferent information from different sensory systems. If these systemsrespond best to change (and velocity information has been reckoned as crucial for balance; Jekaet al., 2004), then static posture would be a barren sensory condition. The argument forproactive, or purposeful, control of sway reasons that the body introduces sway to reap thebenefit of rich afferent information from a constantly changing sensory environment.The work of Gatev et al. (1999) supports the notion of purposeful sway. They showedthat lateral gastrocnemius activity was positively correlated with motion of the center of massand center of pressure during quiet standing, and that the muscle activity lead the motion, ratherthan reacting to it. This lead was longest in a full sensory condition, and shortest in an eyesclosed, Romberg-stance condition. In short, their study showed a feedforward balancemechanism at work, where the body predicted the upcoming anteroposterior sway (upcomingload on the muscle) and activated the muscle to counter it. It was as if the body had planned tosway all along: Gatev called this ‘exploratory behaviour of quiet stance’. Such behaviour is alsoseen by Loram and Lakie (2002a&b), and is further supported by the results of this thesis.As found by Gatev, this study found a lower-leg EMG lead over upper-body behaviour.In Gatev’s experiment it was CoM displacement; in this case, it is head accelerations. Thisleading relationship was true in all muscles and conditions for which timing of maximums andminimums could be determined. Gatev et al. found that EMG in lateral gastrocnemius lead CoMdisplacement by 260-350ms overall. In this study, the anterior and posterior canal lead times for54the first-occurring peak on a hard surface were estimated at —‘1 15-13 5ms in soleus, supportingGatev’s results but showing a shorter lead. This difference may be due to the measures used inthe coherence analysis: Gatev used displacement, whereas we used acceleration. The muscles arealso different.Active systems in standing balanceEither reactive or proactive, coherence in CoM-head acceleration and EMG-headacceleration are mostly confined below 7Hz. Having established that the mechanical strategy instanding under both HS conditions is likely the ankle strategy, we know that the invertedpendulum model dominates in this region. If the inverted pendulum were entirely true, however,the AP and ML force plate/head acceleration coherence would be perfectly coherent (at 1) withinthe frequency bands of quiet stance. This is evidently not the case. Overall lowering of coherencecan be caused by slop in the mechanical linkage through joint flexibility and elasticity, but thereare also defined troughs in the coherence plots. The force-plate/head acceleration coherencefound is probably the result of the mechanical linkage from the foot (where the forces occur) tothe head in the system, all below 7Hz. It is hypothesized that the troughs in coherence are carvedout by active systems interfering in the propagation of forces through the mechanical linkage.This active input functions to reduce coherence in certain frequency bands. This hypothesis issupported by the finding that, in all force/head and EMG/head coherence analyses, conditionsincluding vision had significantly lower coherence under 1Hz (as found by DOC tests). Thissuggests that vision is an active system in standing balance in this region, which aligns withreports that vision contributes most to standing balance at frequencies below 1Hz (Diener &Dichgans, 1988).55The higher-frequency gaps in coherence could be the result of other active systems. It isdifficult to pinpoint exactly where these gaps lie, as their locations differ in each of theinvestigated relationships (although they appear similar in similar planes, AP vs. ML).Fitzpatrick et al. (1992) argued that ankle proprioceptive reflexes act to stabilize the bodybetween 1 and 5Hz, which could cause some of the observed gaps in coherence, perhaps the onesaround 2-3Hz and 4Hz. Fitzpatrick’s experiments used 1-10Hz continuous random perturbations,so his results transfer to reactive control of quiet standing balance, whióh may be part but not allof the story as previously discussed.Another gap is likely caused by active stabilization of the head in stance, as head andbody are thought to be controlled independently during both ankle and hip strategies (Nashner etal., 1988). This active system would decrease coherence in both the EMG-head acceleration andCoM-head acceleration relationships, reflecting the decoupling of EMG and CoM from the headat certain frequencies. The main frequency trough in the EMG-head acceleration coherence isbetween 2-3.5Hz, and parts of that frequency range appear in coherence gaps in the CoM-headacceleration data, as well, leading to the consideration of this region for active head stabilization.This coherence gap is proposed to be of vestibular origin, for two reasons. Firstly, the gap isapparent in the ECSS condition when visual and somatosensory inputs are not reliable andtherefore may not cause a gap. Secondly, head stabilization is one of the proposed main roles ofthe vestibular system in standing balance.The role of the vestibular system in standing balanceIn the beginning of this thesis, four potential purposes of the vestibular system in quietstanding balance were laid out. The results of this thesis do not point to one in particular, ascaution must be taken when interpreting head acceleration results as a direct indication of the56role or importance of the vestibular system. The safest of these interpretations is likely in angularhead acceleration RMS, which alludes to how much acceleration the canals are experiencing.The only difference in angular head acceleration RMS between conditions was in ECSS, themost vestibular-dependent sensory condition, when RMS increased significantly in all directions.There were no main effects of visual or somatosensory (surface) condition; therefore it could bethat angular head acceleration RMS is dependent on the relative importance of the vestibularsystem in the sensory condition.It has been determined that the SCCs are physiologically able to detect the smaller headaccelerations occurring during the other sensory conditions (EOHS, ECHS and EOSS). In themost vestibular-dependent condition, however, head accelerations are allowed to increasesignificantly even though they are detectable when smaller. This could suggest that the vestibularsystem requires more stimulation through higher accelerations in order to participate directly inbalance. This, in turn, seems to support two views. The first is of the vestibular system as“backup” system, which does not create balance responses of its own unless a ‘failsafe’ thresholdis exceeded (which is higher than the perceptual threshold, and according to absolute maximumestimates could be around 3.5-4.0 rad/s2). The second is that the vestibular system is alwaysinvolved in balance through the vestibulospinal reflex, but requires a large input (acceleration) tocreate a large output (reflex). Studies using electrical vestibular stimulation would support thisview, as responses increase with increasing stimulus intensity (Fitzpatrick & Day, 2004;Fitzpatrick et al., 1994). At least small areas of significant coherence were found in this studybetween lower leg EMG and head accelerations in the direction of the SCCs, under allconditions. If those accelerations are proportional to vestibular inputs, then this would suggestthat the vestibular system is involved in balance to different degrees under all conditions,57possibly through the vestibulospinal reflex. The DOC tests show that coherence increases fromEOHS through ECHS and EOSS, and is highest in ECSS as hypothesized. This confirms that thevestibular contribution to balance increases as other sensory information becomes less reliable,although its role cannot be distinguished.These views do not exclude the possibility that the vestibular system acts as a ‘quietpartner’ comparator in lower-acceleration situations, as supposed by Nashner et al. (1982). Inunstable ECSS conditions, it could be that head accelerations are allowed to become significantlyhigher as the limits of balance are reached and the CNS waits for visual and somatosensoryinformation to yield a conclusion against which to compare vestibular information. If they areinconclusive, the vestibular backup response can be executed. If some sway is proactive, ashypothesized previously, then this could add another level to the vestibular comparisons. Thebalance system could use an efferent copy of the proactive AP drive to the ankle muscles topredict what will happen at the head, where the vestibular system resides. This would help thevestibular system in its possible role as a comparator, as comparing predicted to actual inputswould reinforce the verdict. This comparison would become more important as visual andsomatosensory information becomes less reliable (as in ECSS condition), lending anotherexplanation as to why the EMG-head acceleration coherence increases in more vestibulardependent conditions.It is still possible that the vestibular system is responsible for any or all of these roles andperforms head stabilization in space as well. In this way, the vestibular system could beresponsible for a gap in EMG-head acceleration and CoM-head acceleration coherence, aspreviously discussed.58Limitations of the studyIdeally, investigations involving the vestibular system would be completed usingrecordings directly from afferent and efferent nerves in the vestibular apparatus rootlets or theeighth cranial nerve. Not surprisingly these methods cannot be used in studies involving humansubjects, so vestibular outputs must be inferred (and inferences must be recognized and treatedwith caution). In this study, inputs to the vestibular system were estimated to be proportional toangular head accelerations in the direction of each SCC; in turn, vestibular outputs were assumedto be proportional to these stimuli.The novel nature of this study and its equipment goes hand in hand with many inherentlimitations. The first involves drift in linear acceleration measurements. During the long, 4.5 mmtrials it is likely that gradual head or head and body tilt occurred as part of continual posturaladjustments. The position-related effects of gravity on the linear accelerations cannot beaccurately removed from the data in this experimental setup; therefore, very low-frequencycomponents are present in the linear acceleration data (Figure 7). These components introducemore low-frequency power into the signal’s power spectrum, artificially reducing the medianfrequency of the linear accelerations. The same low-frequency component may skew linear headacceleration RMS, as well, giving an artificially high RIVIS to the linear head accelerationsdespite a removal of the mean prior to processing. Kinematic data were not recorded during thetrials, so this effect (caused by gravity) cannot be accurately removed nor can it be correctlyestimated without further information. The exact frequency content of this position-relatedcomponent is unknown. This is an interesting question for an additional study.It is also probable that there is a low-frequency, position-related component in the forceplate data. Carroll and Freedman (1993) found that quiet standing CoP was not a stationary59stochastic process, as had been previously assumed. Instead, average value and sway variancewere time dependent, indicating travel in the center of CoP over time. The combination ofposition-related drift in both CoP and linear head acceleration data may cause a low-frequencyspike in coherence between the signals that is misleading.Other equipment limitations include intrinsic off-axis sensitivity in the accelerometers,possible drift due to temperature over the long trial time and, of course, potential head gearmovement on the subject. Although the head apparatus was tightened and subjectively felt secureon each subject, parts of the apparatus were non-rigid plastic. This allowed for betteradjustability and fit, but may have permitted some equipment shift during trials. Hair and skinshift over the skull was unavoidable using this experimental setup. By inspection, it appears thatthe accelerometer array was able to pick up very small angular and linear head accelerations, asseen in a sample second of angular and linear acceleration data (Figure 8).When force plate data were recorded, efforts were made to ensure that the subject’s headwas aligned with equivalent force plate axes. The recording of natural head movement wasparamount, however, and instructions needed to be limited. It is inevitable that these axes fell outof alignment during the trials, indeed in some subjects the inferiosuperior (z) axes never lined upat all. The coherence analyses on data collected in equivalent axes must therefore be treated withcare, as these axes may be approximate but not equivalent, per se.60AngularAccelerationradls2LinearAccelerationrn/s20.50.5EOHS Accelerations: Single Representative SubjectFigure 7: Angular and linear accelerations about Reid’s axes for one EOHS trial in a representativesubject over 4.5 minutes. Blue represents x (roll & AP), green is y (pitch & ML), red is z (yaw &IS). Means have been removed. Concurrent bursting in angular and linear accelerations can beseen around 135s. The low-frequency position-related component of linear acceleration can beclearly seen here, especially in ML and AP directions.AngularAccelerationradIs2Linear04Accelerationrn/s200./_—.——1r-\-——%’0.5Time(s)Figure 8: Processed accelerometer data from one second in the middle of one EOHS trial in arepresentative subject. Blue represents x (roll & AP), green is y (pitch & ML), red is z (yaw & IS).Overall means have been removed; reference frame is Reid’s axes.0 50 100 150 200 250Time (s)EOHS Accelerations: One second: in a single subject61The transformation of head acceleration data from Reid’s axes to the directions of theSCCs was an appealing and challenging direction to this study. Recent work has mapped theorientations of the individual canals in the human skull to a tenth of a degree (Della Santina etal., 2005). However, the authors warn against using this data to infer prime directionality of thecanals, as anatomic planes align poorly with these prime directions, and the high inter-subjectvariability of these planes and the canals makes the estimates even worse. Accelerations in thedirections of the SCCs have been calculated here, but are certainly estimates only. Additionally,it should be mentioned that the linear accelerations experienced by the vestibular apparatuscannot be discerned from this study. This is due not only to the previously mentioned limitationsof the linear measurements, but also because no orientation information exists for human otolithorgans, so no transformation could be completed.The use of EMG in our coherence analysis also comes with limitations. Overall lowmuscle activity could lead to a low signal-to-noise ratio, and cross-talk in any of the signalscould create misleading results. Care was taken to prevent cross-talk by appropriate landmarking,but the possibility is recognized. In terms of muscle specificity, it is evident from the coherenceplots that medial gastrocnemius and soleus show higher vestibular connectivity that tibialisanterior. As expected, coherence in tibialis anterior rarely reached significance, even in the mostvestibular-dependent conditions. This suggests a lack of vestibular coupling in TA, yet thisconclusion is not supported by GVS studies (Fitzpatrick et al., 1994; Cathers et al., 2005). It ispossible that coherence was not found in TA because the level of activation of this muscle underthe conditions studied is small relative to its antagonists. This lowers the signal-to-noise ratio andmakes it more difficult to identify a real relationship between the data.62There are a few data sets that would have added to this study. For example, accelerationswere recorded from the head only, which fulfilled the aims of the study but brought up manyother questions. Concurrent relative behaviour of the torso may have answered some of them butwas estimated, not recorded. Concurrent center of pressure under soft surface conditions wouldalso add to the data set. Unfortunately the validity of CoP when measured under foam isquestionable, and therefore CoP data could only be analyzed for EO and EC conditions on thehard surface, limiting its usefulness. Hard surface CoP measures are available in Appendix 4.63ConclusionsThe results of this study have provided a reference for angular and linear headacceleration behaviour during quiet standing sway in healthy humans under different visual andsomatosensory conditions. It has been shown that angular head accelerations are repeatableunder full sensory conditions, and that angular head acceleration RMS is above quoted vestibularthresholds in all tested sensory conditions. Linear head acceleration absolute maximum and RMSvalues tended to match previous reports under similar conditions.The results reinforce that the inverted pendulum model is valid in quiet standing on ahard surface in the sagittal and frontal planes. Independent mechanisms of balance in theseplanes are maintained. The mechanical linkage in both planes confines the CoM-headacceleration and EMG-head acceleration coherence to below 7Hz, under which active systemsare hypothesized to carve out frequency bands of influence. The visual system is suggested tooperate below 1Hz, whereas the somatosensory and vestibular systems are proposed to haveimpact in the 2-4 Hz range.The study also indicates that the vestibular system may play different roles in quietstanding balance as its importance in the maintenance of balance increases, but best supports thetheory of the vestibular system as a head stabilizer and comparator. A third comparable variable(besides vision and somatosensory feedback) is proposed to be a vestibular input prediction,created from an efferent copy of the proactive drive to the ankle flexors and extensors.This thesis study successfully measured the small accelerations occurring at the head inquiet stance. It uncovered limitations that afflicted the methods both in equipment and inanalysis, and thus helped to ascertain the improvements that can be made to future studies in the64area. It is hoped that the achievements of this thesis will promote the use of head accelerographyin quiet stance, a technique that has not yet been fuiiy exploited in humans.65List of referencesAu, A. S., Rowen, K. A., & lies, J. F. (2003). Vestibular actions on back and lower limb musclesduring postural tasks in man. The Journal ofPhysiology, 546(Pt 2), 615-624.Allum, J. H., Zamani, F., Adkin, A. L., & Ernst, A. (2002). Differences between trunk swaycharacteristics on a foam support surface and on the equitest ankle-sway-referenced supportsurface. Gait & Posture, 16(3), 264-270.Amiridis, I. G., Hatzitaki, V., & Arabatzi, F. (2003). Age-induced modifications of staticpostural control in humans. Neuroscience Letters, 350(3), 13 7-140.Amjad, A. M., Halliday, D. M., Rosenberg, J. R., & Conway, B. A. (1997). An extendeddifference of coherence test for comparing and combining several independent coherenceestimates: Theory and application to the study of motor units and physiological tremor.Journal ofNeuroscience Methods, 73(1), 69-79.Angelaki, D. F., Shaikh, A. G., Green, A. M., & Dickman, J. D. (2004). Neurons computeinternal models of the physical laws of motion. Nature, 430(6999), 560-564.Ardic, F. N., Latt, L. D., & Redfern, M. S. (2000). Paraspinal muscle response to electricalvestibular stimulation. Acta Oto-Laryngologica, 120(1), 39-46.Benson, A. J., Hutt, E. C., & Brown, S. F. (1989). Thresholds forthe perception of whole bodyangular movement about a vertical axis. Aviation, Space, and Environmental Medicine,60(3), 205-2 13.Black, F. 0., Shupert, C. L., Horak, F. B., & Nashner, L. M. (1988). Abnormal postural controlassociated with peripheral vestibular disorders. Progress in Brain Research, 76, 263-275.Blouin, J. S., Siegmund, G. P., & Timothy Inglis, J. (2007). Interaction between acoustic startleand habituated neck postural responses in seated subjects. Journal ofApplied Physiology,102(4), 1574-1586.Blumenfeld, H. (2002). In Blumenfeld H. (Ed.), Neuroanatomy through clinical cases (1st ed.).Sunderland, MA: Sinauer Associates, Inc.Britton, T. C., Day, B. L., Brown, P., Rothwell, J. C., Thompson, P. D., & Marsden, C. D.(1993). Postural electromyographic responses in the arm and leg following galvanicvestibular stimulation in man. Experimental Brain Research, 94(1), 143-151.Carpenter, M. B. (1988). Vestibular nuclei: Afferent and efferent projections. Progress in BrainResearch, 76, 5-15.66Carpenter, M. G., Frank, J. S., Winter, D. A., & Peysar, G. W. (2001). Sampling duration effectson centre of pressure summary measures. Gait & Posture, 13(1), 3 5-40.Carroll, J. P., & Freedman, W. (1993). Nonstationary properties of postural sway. Journal ofBiomechanics, 26(4-5), 409-416.Cathers, I., Day, B. L., & Fitzpatrick, R. C. (2005). Otolith and canal reflexes in human standing.The Journal ofPhysiology, 563(Pt 1), 229-234.Clark, B., & Stewart, J. D. (1970). Thresholds for the perception of angular acceleration aboutthe three major body axes. Acta Oto-Laryngologica, 69(4), 231-238.Cullen, K. E., & Minor, L. B. (2002). Semicircular canal afferents similarly encode active andpassive head-on-body rotations: Implications for the role of vestibular efference. TheJournal ofNeuroscience: The Official Journal ofthe Societyfor Neuroscience, 22(11),RC226.Dakin, C. J., Son, G. M., Inglis, J. T., & Blouin, J. 5. (2007). Frequency response of humanvestibular reflexes characterized by stochastic stimuli. The Journal ofPhysiology, 583(Pt 3),1117-1 127.Day, B. L., & Fitzpatrick, R. C. (2005). The vestibular system. Current Biology: CB, 15(15),R583-6.Della Santina, C. C., Potyagaylo, V., Migliaccio, A. A., Minor, L. B., & Carey, J. P. (2005).Orientation of human semicircular canals measured by three-dimensional multiplanar CTreconstruction. Journal ofthe Association for Research in Otolaryngology : JAR 0, 6(3),191-206.Diener, H. C., & Dichgans, J. (1988). On the role of vestibular, visual and somatosensoryinformation for dynamic postural control in humans. Progress in Brain Research, 76, 253-262.Dietz, V. (1992). Human neuronal control of automatic functional movements: Interactionbetween central programs and afferent input. Physiological Reviews, 72(1), 33-69.Drake, R. L., Vogl, W., & Mitchell, A. W. M. (2005). Head and neck. In R. L. Drake, W. Vogl& A. W. M. Mitchell (Eds.), Gray’s anatomyfor students (1st ed.,pp. 747--1033). Toronto:Elsevier Inc.Easton, R. D., Greene, A. J., DiZio, P., & Lackner, J. R. (1998). Auditory cues for orientationand postural control in sighted and congenitally blind people. Experimental Brain Research,118(4), 541-550.Farmer, S. F., Bremner, F. D., Halliday, D. M., Rosenberg, J. R., & Stephens, J. A. (1993). Thefrequency content of common synaptic inputs to motoneurones studied during voluntaryisometric contraction in man. The Journal ofPhysiology, 470, 127-155.67Fitzpatrick, R. C., Gorman, R. B., Burke, D., & Gandevia, S. C. (1992). Postural proprioceptivereflexes in standing human subjects: Bandwidth of response and transmissioncharacteristics. The Journal ofPhysiology, 458, 69-83.Fitzpatrick, R.C., & McCloskey, D. I. (1994). Proprioceptive, visual and vestibular thresholds forthe perception of sway during standing in humans. The Journal ofPhysiology, 478 (Pt 1),173-186.Fitzpatrick, R.C., Burke, D., & Gandevia, S. C. (1994). Task-dependent reflex responses andmovement illusions evoked by galvanic vestibular stimulation in standing humans. TheJournal ofPhysiology, 478 (Pt 2), 363-372.Fitzpatrick, R. C., & Day, B. L. (2004). Probing the human vestibular system with galvanicstimulation. Journal ofApplied Physiology, 96(6), 2301-2316.Fitzpatrick, R. C., Butler, J. E., & Day, B. L. (2006). Resolving head rotation for humanbipedalism. Current Biology: CB, 16(15), 1509-15 14.Gage, W. H., Winter, D. A., Frank, J. S., & Adkin, A. L. (2004). Kinematic and kinetic validityof the inverted pendulum model in quiet standing. Gait & Posture, 19(2), 124-132.Gatev, P., Thomas, S., Kepple, T., & Hallett, M. (1999). Feedforward ankle strategy of balanceduring quiet stance in adults. The Journal ofPhysiology, 514 (Pt 3)(Pt 3), 9 15-928.Goldberg, J. M. (2000). Afferent diversity and the organization of central vestibular pathways.Experimental Brain Research, 130(3), 277-297.Goldberg, J. M., Smith, C. E., & Fernandez, C. (1984). Relation between discharge regularityand responses to externally applied galvanic currents in vestibular nerve afferents of thesquirrel monkey. Journal ofNeurophysiology, 51(6), 1236-1256.Halliday, D. M., & Rosenberg, J. R. (1999). Time and frequency domain analysis of spike trainand time series data. In U. Windhorst, & H. Johansson (Eds.), Modern techniques inneuroscience research(pp.503-543) Springer-Verlag.Halliday, D. M., Rosenberg, J. R., Amjad, A. M., Breeze, P., Conway, B. A., & Farmer, S. F.(1995). A framework for the analysis of mixed time series/point process data - theory andapplication to the study of physiological tremor, single motor unit discharges andelectromyograms. Prog Biophys Mol Biol, 64, 23 7-278.Highstein, S. M., Rabbitt, R. D., Holstein, G. R., & Boyle, R. D. (2005). Determinants of spatialand temporal coding by semicircular canal afferents. Journal ofNeurophysiology, 93(5),2359-2370.Horak, F. B., & MacPherson, J. M. (1996). Postural orientation and equilibrium. In J. L. Smith(Ed.), Exercise: Regulation and integration ofmultiple systems (1st ed.,pp.255-292). NewYork: Oxford University Press.68Horak, F. B., Shupert, C. L., Dietz, V., & Horstmann, G. (1994). Vestibular and somatosensorycontributions to responses to head and body displacements in stance. Experimental BrainResearch, 100(1), 93-106.Ishizaki, H., Pyykko, I., Aalto, H., & Starck, J. (1991). Repeatability and effect of instruction ofbody sway. Acta Oto-Laryngologica. Supplementum, 481, 589-592.Jeka, 3., Kiemel, T., Creath, R., Horak, F., & Peterka, R. (2004). Controlling human uprightposture: Velocity information is more accurate than position or acceleration. Journal ofNeurophysiology, 92(4), 23 68-2379.Jeka, J., & Lackner, J. R. (1994). Fingertip contact influences human postural control.Experimental Brain Research, 100(3), 495-502.Kelly, J. W., Loomis, J. M., & Beall, A. C. (2005). The importance of perceived relative motionin the control of posture. Experimental Brain Research, 161(3), 285-292.Kennedy, P. M., Cresswell, A. G., Chua, R., & Inglis, J. T. (2004). Vestibulospinal influences onlower limb motoneurons. Canadian Journal ofPhysiology and Pharmacology, 82(8-9), 675-681.Keshner, E. A., & Dhaher, Y. (2008). Characterizing head motion in three planes duringcombined visual and base of support disturbances in healthy and visually sensitive subjects.Gait & Posture, 28(1), 127-134.King, A. I. (1993). Progress of research on impact biomechanics. Journal ofBiomechanicalEngineering, 115(4B), 582-587.Kitahara, M. (1959). Study on the acceleration of the head during human equilibrium movement.A preliminary report. Practica Otologica (Kyoto), 52, 1044-1064.Kitahara, M. (1965). Acceleration registrography: A new method of examinations concernedwith the labyrinthine righting reflex. The Annals ofOtology, Rhinology, and Laryngology,74, 203-2 14.Klinke, R., & Galley, N. (1974). Efferent innervation of vestibular and auditory receptors.Physiological Reviews, 54(2), 3 16-357.Lafond, D., Corriveau, H., & Prince, F. (2004). Postural control mechanisms during quietstanding in patients with diabetic sensory neuropathy. Diabetes Care, 27(1), 173-178.Latt, M. D., Menz, H. B., Fung, V. S., & Lord, S. R. (2008). Walking speed, cadence and steplength are selected to optimize the stability of head and pelvis accelerations. ExperimentalBrain Research, 184(2), 201-209.Lee Son, G. M., Blouin, J. S., & Inglis, J. T. (2005). Assessing the modulations investibulospinal sensitivity during quiet standing using galvanic vestibular stimulation.Neuroscience, Washington, DC.69Loram, I. D., & Lakie, M. (2002a). Direct measurement of human ankle stiffness during quietstanding: The intrinsic mechanical stiffness is insufficient for stability. The Journal ofPhysiology, 545(Pt3), 1041-1053.Loram, I. D., & Lakie, M. (2002b). Human balancing of an inverted pendulum: Position controlby small, ballistic-like, throw and catch movements. The Journal ofPhysiology, 540(Pt 3),1111-1124.Lund, S., & Broberg, C. (1983). Effects of different head positions on postural sway in maninduced by a reproducible vestibular error signal. Acta Physiologica Scandinavica, 117(2),307-309.Lysakowski, A., & Goldberg, J. M. (2004). Morphophysiology of the vestibular periphery. In S.M. Highstein, R. R. Fay & A. N. Popper (Eds.), The vestibular system (1st ed.,pp.57-152).New York: Springer.Lysakowski, A., Minor, L. B., Fernandez, C., & Goldberg, J. M. (1995). Physiologicalidentification of morphologically distinct afferent classes innervating the cristae ampullaresof the squirrel monkey. Journal ofNeurophysiology, 73(3), 1270-128 1.MacDougall, H. G., Moore, S. T., Curthoys, I. S., & Black, F. 0. (2006). Modeling posturalinstability with galvanic vestibular stimulation. Experimental Brain Research, 172(2), 208-220.Macpherson, J. M., & Inglis, J. T. (1993). Stance and balance following bilaterallabyrinthectomy. Progress in Brain Research, 97, 219-228.Matsubara, H., Kitahara, M., Saito, H., Kitajima, K., Takeda, T., Yazawa, Y., et al. (1983).Correlation between head movement and body sway. Agressologie, 24(2), 93-94.Mcllroy, W. E., & Maki, B. E. (1997). Preferred placement of the feet during quiet stance:Development of a standardized foot placement for balance testing. Clinical Biomechanics,12(1), 66-70.McNevin, N. H., & Wulf, G. (2002). Attentional focus on supra-postural tasks affects posturalcontrol. Human Movement Science, 21(2), 187-202.Meiry, J. L. (1966). The vestibular system and human dynamic space orientation. NASA CR628. NASA Contractor Report.NASA CR. United States. National Aeronautics and SpaceAdministration, 1-192.Miyoshi, T., Hiwatashi, S., & Tamada, A. (1983). Correlation between body sway and head tilt.Agressologie, 24(2), 9 1-92.Nagano, A., Yoshioka, S., Hay, D. C., & Fukashiro, S. (2006). Light finger touchon the upperlegs reduces postural sway during quasi-static standing. Motor Control, 10(4), 348-35 8.70Nashner, L. M. (1971). A model describing vestibular detection of body sway motion. Acta OtoLaryngologica, 72(6), 429-436.Nashner, L. M. (1976). Adapting reflexes controlling the human posture. Experimental BrainResearch, 26(1), 59-72.Nashner, L. M., Black, F. 0., & Wall, C.,3rd. (1982). Adaptation to altered support and visualconditions during stance: Patients with vestibular deficits. The Journal ofNeuroscience:The Official Journal ofthe Societyfor Neuroscience, 2(5), 53 6-544.Nashner, L. M., Shupert, C. L., & Horak, F. B. (1988). Head-trunk movement coordination in thestanding posture. Progress in Brain Research, 76, 243-251.Nashner, L. M., Shupert, C. L., Horak, F. B., & Black, F. 0. (1989). Organization of posturecontrols: An analysis of sensory and mechanical constraints. Progress in Brain Research,80, 411-8; discussion 3 95-7.National Aeronautics and Space Administration. (1978). Anthropometric source book (Ref. Pubi.1024 ed.). Hanover, MD.: National Aeronautics and Space Administration Scientific andTechnical Information Office.Nelson, J. G. (1968). Effect of water immersion and body position upon perception of thegravitational vertical. Aerospace Medicine, 39(8), 806-811.Padgaonkar, A. J., Krieger, K. W., & King, A. I. (1975). Measurement of angular acceleration ofa rigid body using linear accelerometers. iAppl.Mech., (42), 552-556.Pavlik, A. E., Inglis, J. T., Lauk, M., Oddsson, L., & Collins, J. J. (1999). The effects ofstochastic galvanic vestibular stimulation on human postural sway. Experimental BrainResearch, 124(3), 273-280.Plotnik, M., Marlinski, V., & Goldberg, J. M. (2005). Efferent-mediated fluctuations investibular nerve discharge: A novel, positive-feedback mechanism of efferent control.Journal ofthe Associationfor Research in Otolaiyngology . JARO, 6(4), 311-323.Rabbitt, R. D., Damiano, E. R., & Grant, J. W. (2004). Biomechanics of the semicircular canalsand otolith organs. In S. M. Highstein, R. R. Fay & A. N. Popper (Eds.), The vestibularsystem (1st ed.,pp.153-201). New York: Springer.Rosenberg, J. R., Amjad, A. M., Breeze, P., Brillinger, D. R., & Halliday, D. M. (1989). Thefourier approach to the identification of functional coupling between neuronal spike trains.Progress in Biophysics and Molecular Biology, 53(1), 1-31.Roskell, M. G., Dakin, C. J., McFadyen, B. J., Chua, R., Blouin, J. S., & Inglis,J. T. (2007).Vestibulomotor reflexes evoked by stochastic vestibular stimulation in walking humans.Neuroscience, San Diego.71Ross, M. D. (2003). The evolution of concepts of vestibular peripheral information processing:Toward the dynamic, adaptive, parallel processing macular model. Acta Oto-Laryngologica,123(7), 784-794.Ruckenstein, M. J. (2004). In Ruckenstein M. J. (Ed.), Comprehensive review ofotolaryngology(1st ed.). New York: Saunders.Sakaguchi, M., Taguchi, K., Ishiyama, T., Netsu, K., & Sato, K. (1995). Relationship betweenhead sway and center of foot pressure sway. Auris, Nasus, Larynx, 22(3), 15 1-157.Scinicariello, A. P., Inglis, J. T., & Collins, J. J. (2002). The effects of stochastic monopolargalvanic vestibular stimulation on human postural sway. Journal of Vestibular Research:Equilibrium & Orientation, 12(2-3), 77-85.Shumway-Cook, A., & Woollacott, M. (2000). Attentional demands andpostural control: Theeffect of sensory context. The Journals ofGerontology.Series A, Biological Sciences andMedical Sciences, 55(1), M10-6.Simoneau, G. G., Ulbrecht, J. S., Derr, J. A., Becker, M. B., & Cavanagh, P. R. (1994). Posturalinstability in patients with diabetic sensory neuropathy. Diabetes Care, 17(12), 14 11-1421.Soames, R. W., & Atha, J. (1982). The spectral characteristics of postural sway behaviour.European Journal ofApplied Physiology and Occupational Physiology, 49(2), 169-177.Spoor, F., Wood, B., & Zonneveld, F. (1994). Implications of early hominid labyrinthinemorphology for evolution of human bipedal locomotion. Nature, 369(6482), 645-648.Stoffregen, T. A., Bardy, B. G., Bonnet, C. T., Hove, P., & Oullier, 0. (2007). Postural sway andthe frequency of horizontal eye movements. Motor Control, 11(1), 86-102.Tsujikawa, 0. (1966). Acceleration registrography in standing position - results in normal humanbeings. Bull. Yamaguchi Med. School, 13(3), 13 1-147.Uchino, Y., Sasaki, M., Sato, H., Bai, R., & Kawamoto, E. (2005). Otolith and canal integrationon single vestibular neurons in cats. Experimental Brain Research, 164(3), 271-285.Vuillerme, N., & Nafati, G. (2007). How attentional focus on body sway affects postural controlduring quiet standing. Psychological Research, 71(2), 192-200.Weeks, D. L., Forget, R., Mouchnino, L., Gravel, D., & Bourbonnais, D. (2003). Interactionbetween attention demanding motor and cognitive tasks and static postural stability.Gerontology, 49(4), 225-232.Weir, J. P. (2005). Quantifying test-retest reliability using the intraclass correlation coefficientand the SEM. Journal ofStrength and Conditioning Research /National Strength &Conditioning Association, 19(1), 231-240.72Welgampola, M. S., & Colebatch, J. G. (2001). Vestibulospinal reflexes: Quantitative effects ofsensory feedback and postural task. Experimental Brain Research, 139(3), 345-353.Welgampola, M. S., & Colebatch, J. G. (2002). Selective effects of ageing on vestibulardependent lower limb responses following galvanic stimulation. Clinical Neurophysiology:Official Journal ofthe International Federation ofClinical Neurophysiology, 113(4), 528-534.Winter, D. A. (1995). In Winter D. A. (Ed.), A.B.C. (anatomy, biomechanics and control) ofbalance during standing and walking (1st ed.). Waterloo, ON: Waterloo Biomechanics.Winter, D. A., Prince, F., Frank, J. S., Powell, C., & Zabjek, K. F. (1996). Unified theoryregarding A/P and M/L balance in quiet stance. Journal ofNeurophysiology, 75(6), 2334-2343.Winter, D. A., Patla, A. E., Prince, F., Ishac, M., & Gielo-Perczak, K. (1998). Stiffness control ofbalance in quiet standing. Journal ofNeurophysiology, 80(3), 1211-1221.Winter, D. A., Patla, A. E., Ishac, M., & Gage, W. H. (2003). Motor mechanisms of balanceduring quiet standing. Journal ofElectromyography and Kinesiology: Official Journal ofthe International Society ofElectrophysiological Kinesiology, 13(1), 49-56.Yakushin, S., Dai, M., Suzuki, J., Raphan, T., & Cohen, B. (1995). Semicircular canalcontributions to the three-dimensional vestibuloocular reflex: A model-based approach.Journal ofNeurophysiology, 74(6), 2722-2738.Yardley, L., Gardner, M., Leadbetter, A., & Lavie, N. (1999). Effect of articulatory and mentaltasks on postural control. Neuroreport, 10(2), 215-219.Yoganandan, N., Zhang, J., Pintar, F. A., & King Liu, Y. (2006). Lightweight low-profile nine-accelerometer package to obtain head angular accelerations in short-duration impacts.Journal ofBiomechanics, 39(7), 1347-1354.Yoneda, S., & Tokumasu, K. (1986). Frequency analysis of body sway in the upright posture.statistical study in cases of peripheral vestibular disease. Acta Oto-Laryngologica, 102(1-2),87-92.Zok, M., Mazza, C., & Cappozzo, A. (2008). Should the instructions issued to the subject intraditional static posturography be standardised? Medical Engineering & Physics,30(7):9 13-6.73Appendix 1: Tables of control trial dataI Control 1 I Exp 1 Control 2 MeanROLL(x)Absolute Max 1.96+1-1.68 1.88÷1-1.09 2.23+1-1.42 rad/s2Overall RMS 0.27+1-0.06 0.27+1-0.06 0.28÷1-0.06 0.27÷1-0.06 rad/s2Median Freq 9.04 -4-/- 2.41 8.45+1-2.50 8.01-‘-I-2.28 8.50+1-2.36 Hz95% CIF 21.45÷1-2.66 21.35+1-2.75 20.21÷1-1.86 HzPITCH(v)Absolute Max 2.09+1-1.39 2.42 -i-f- 0.90 2.64÷1-0.92 rad/s2Overall RMS 0.28+1-0.05 0.29+1-0.05 0.32+1-0.06 0.30÷1-0.06 rad/s2Median Freq 4.05+1-0.90 3.95÷1-0.80 3.82+1-0.73 3.94÷1-0.80 Hz95% CIF 18.68+1-3.79 17.66+1-3.31 16.49+1-2.20 HzAbsolute Max 2.28+1-2.21 2.21+1-1.61 2.68-I-I-1.32 rad/s2Overall RMS 0.28+1-0.06 0.29+1-0.06 0.32+1-0.07 0.29÷1-0.06 rad/s2Median Freq 4.84+1-0.68 4.62+1-0.50 4.53÷1-0.69 4.66+1-0.62 Hz95% CIF 15.54 +1- 2.21 15.56 +1- 2.39 15.61+1-2.40 HzTable 5: Results for angular accieration of the head in roll, pitch and yaw during threecontrol trials in the eyes-open, hard-surface condition. Mean +1- standard deviation betweensubjects. Final column shows overall means and standard deviations across all trials, if an ICC wasperformed and showed reliability.74YAW(z)____________IControl 1 Exp 1 Control 2 MeanAP (x)Absolute Max 237+1-1.17 2.57+1-1.07 2.79+1-1.28 rn/s2Overall RMS 0.19+1-0.11 0.24÷1-0.09 0.20+1-0.08 0.21+1-0.09 rn/s2Median Freq 0.03+1-0.04 0.01+1-0.02 0.01+1-0.01 0.02+1-0.03 Hz95% CIF 2.09+1-1.98 1.17+1-1.58 2.20+1-2.37 Hz___________ML(y)Absolute Max 0.66÷1-0.19 0.73÷1-0.25 0.90+1-0.31 rn/s2Overall RMS 0.09+1-0.03 0.11+1- 0.05 0.11 +1-0.04 rn/s2Median Freq 0.02+1-0.02 0.05÷1-0.06 0.06+1-0.07 Hz95% CIF 12.56 -i-/- 3.03 9.93-‘-I-5.74 10.08+1-5.28 HzAbsolute Max 9.79+1-0.14 9.79÷1-0.18 9.78+1-0.21 rn/s2Overall RMS 0.05+1-0.05 0.06+1-0.04 0.06+1-0.04 0.06+1-0.04 rn/s2Median Freq 1.08+1-2.92 0.74+1-2.14 0.42+1-1.05 Hz95% CIF 23.82+1-11.12 17.68+1-11.26 17.71+/-11.53 HzTable 6: Results for linear accieration of the head in the anteroposterior (AP), mediolateral (ML)and inferiosuperior (IS) directions during three control trials in the eyes-open, hard-surfacecondition. Mean +1- standard deviation between subjects. Final column shows overall means andstandard deviations across all trials, unless the ICC was performed and did not show consistencybetween data sets.75IS (z)Appendix 2: Differences in acceleration input to each SCCA series of 1-way, 3-level ANOVAs were run on angular acceleration RMS and medianfrequency in the directions of the 3 SCCs, in order to identif’ any differences in angularacceleration felt by each canal. A summary of the information received by all canals can befound in Table 4 on page 40. The tests were completed to a p-value of 0.00625, representing a pvalue of 0.05 with a Bonferroni correction factor for multiple comparisons of 8. If significant,post-hoc Tukey tests were performed to the same significance value to discover wheredifferences lay. Statistically significant differences appeared in RMS in the ECSS condition only,showing the horizontal canal angular acceleration RMS to be smaller than the posterior canalvalues (F(2,20)=8.56, p=O.OO22). There was one difference in median frequency: in the ECHScondition, horizontal canal median frequency was significantly higher than the anterior’s(F(2,20)=1 1.4 16, p=0.000542).Linear head acelerations in the directions of the semicircular canals were not analyzed.76Appendix 3: Coherence between head accelerations andEMG dataCoherence analysis was run between angular head accelerations in the planes of the rightsemi-circular canals and EMG data from 3 muscles bilaterally: medial gastrocnemius, soleus,and TA. Results can be seen in Figure 9 for the anterior canal, Figure 10 for the horizontal canal,and Figure 11 for the posterior canal. TCD plots are shown in Figure 12.77Coherence: Head accel in direction of right anteriorcanal vs. muscleLL0.20.2Figure 9: Coherence between anterior canal accelerationsand EMG. Horizontal lines denote 95%CI. The first plots are coherence: EOHS, ECHS, EOSS, and ECSS incolumns as labeled. Thesecond plot shows a 4-way DOC test between all conditions.Pooied Coherrnce EstimtebOC (4way)ECHS EOSS ECSS0.2R0.2iJbUJi&0,2FtI00C0.2RL0F250 20 400 20 400 20 40780CCoherence: Head accel in directionof right horizontalcanal vs. muscleFigure 10: Coherence betweenright horizontal canal accelerationsand EMG. Horizontallinesdenote 95% CI. The first columnis coherence: EOHS, ECHS,EOSS, and ECSS in columnsaslabeled. The second column showsthe results of a 4-way DOC testbetween all conditions.EOHSPooled Coherence EstrnateECHS LOSSLi. L...ECSS1.LVI0DOC (4way)O21..02R0.20.2K02L:fr4As___0ii . LLL1iLl JJ.. L..0 20400 20 40Freq (Hz)0 20 4079Coherence Head accel indirection of right posterior canalvs. muscleFigure 11: Coherence betweenposterior canal accelerationsand EMG. Horizontal lines denote95% CI. The first column is coherence:EOHS, ECHS, EOSS, and ECSS incolumns as labeled. Thesecond column shows the results ofa 4-way DOC test between all conditions.0.2KEOHSPooecJ Coherence EstimateECH$ E0SSIIDOC (4way)EC$$-14j&L—-0.2L0.2R0.2L0.2K0.2La0)CCCV.’J’hJ’0.1Li. .250 20 40ALj.ü Io 20 4020 40Freq(Hz) ..“..- —80C0CC0C01)0Figure 12: TCD plots betweencanal accelerations andEMG under four sensory conditions.Blacklines are right muscles; grey,left. Blue parts of the plot do notcross the boundaries of significance.MGsoTime Cumulant Density:Head Acceleration &EMGEOHS ECHSEOSS ECSS-H---•+MG -—*1J_so—--—*——--‘:_LTAF—-—f-—MG-4-so.01-.01-1 0Time(s)181Appendix 4: Center of pressuremeasuresCOP RIVIS was repeatable in both AP and ML (ICC> 0.653),but was not at all in medianfrequency (ICC <0.379 for AP and ML).Control 1 Exp 1 Control 2MeanAnteroposterior (x)Absolute Max 23.62+1-12.71 24.41÷1-9.57 31.19+1-13.64 mmOverall RMS 6.68÷1-1.99 6.78+1-1.82 7.04+1-2.51 6.83+1-2.89 mmMedian Freq 0.04+1-0.02 0.03+1-0.02 0.05+1- 0.03 Hz95% CIF 0.54+1-0.21 0.58+1-0.23 0.53+1-0.21 HzAbsolute Max 15.23+1-6.41 20.09+1-11.09 20.11+1-9.43 mmOverall RMS 4.55+1-1.58 5.34+1-1.52 4.91+1-1.77 4.93+1-3.23 mmMedian Freq 0.05+1-0.02 0.06+1-0.05 0.10+1-0.06 Hz95% CIF 0.69+1-0.26 0.60+1-0.17 0.72+1-0.23 HzTable 7: Results for COP displacement of in the anteroposterior(AP) and mediolateral (ML)directions during three control trials in the EOHScondition. Mean +1- standard deviation betweensubjects. Final column shows overall means and standarddeviations across all trials if theICC wasperformed and showed consistency between datasets.Expi Exp2EOHS ECHSAnteroposterior (x)Absolute Max 24.41 4-/- 9.57 28.51+1-13.37 mmOverall RMS 6.78+1-1.82 8.57+1-1.97 mmMedian Freq 0.03+1-0.02 0.06÷1-0.05 Hz95%CIF 0.58+1-0.23 0.63+1-0.18 HzAbsolute Max 20.09+1-11.09 23.76+1-6.74 mmOverall RMS 5.34+1-1.52 5.96+1-1.40 mmMedian Freq 0.06÷1-0.05 0.15+1-0.08 Hz95% CIF 0.60+1-0.17 0.86+1-0.31 HzTable 8: Results for COP displacement of in the anteroposterior(AP) and mediolateral (ML)directions during two trials: EOHS andECHS. Mean +1- standard deviation between subjects.82Mediolateral (y)Mediolateral (y)Appendix 5: Research Ethics Board Certificate of ApprovalThe Urwersltyof Bntinh CofurrntiaOffIce of Research deivicesClinical Research Eliirs hoard— Room 210, 828 VWst 11th Anemia, Vancower 80 V5Z fIBETHICS CERTIFICATE OF FULL BOARD APPROVALRR4CIPAL INVESTIGATOR NST[fUTIONI DEPARTMENT: JBC CRES NUMBER:ean-Sabaebun Sleds jJoClEducabnml*nnao Ktatlca [107-03119rJSTITUTION(S) WHERE RESEARCH WILL SE CARRIED OUT:had*tthn ISC Vancorwer (euckulea USC tleepeel)tberb.tia.swrneaierench iI becnduceed:IAD-INVESTIGATDR(S):lartin E HdrouoTimothy Inglishris Oakinanlol Mangomoo Chualelanie 0. RoskellPDNSDRING AGENCIES:Natural Sciences and Engineedng Research Council of Canada (NSERC)-“Netjroyh,’elrioq,’ of human neck and lower limb motoneuruns”RDJECT TITLE:ourophyalology of human rack art lower limba motoneuronaHE CURRENT USC CRES APPROVAL FDR THIS STUDY EXPIRES: April 22,2009he full UBC Clinical Research Ethics Beard has reviewed the above described research project, including asoociated documentation onled below, aod nods the research projacceptabla an ethical grnuods for research ionafrmna human sabjecta and hereby granta approval.ES FULL SDARD MEETING REVIEW DATE:prIl 22, 20tDCUMENTS INCLUDED IN THIS APPROVAL: ‘ATE DOCUMENTS APPROVED:,munpreWaimIvadonI Oweretocol:eaearch Protocol Version 2 May 13, 2006onsmtu Farms:lay 21,21118onsurt form Version 2 May 13, 2006dverllsaments:dhmrtiaement Version 2 May 13, 2006‘)oestiersnaire. Daeatioonaire Cover Letter. Tears:Eplepsy qoestionnaie Version 1 May 13, 200aUttsI inlr..F\ilLri’uIn respect of clinical trials:I. The membership of this Research Ethics hoseS complies with the membership requirements for Research Ethics Boards defined in Division S of the Food aed DregRegulations.2. The Research Ethics hoard carries oct its functions in a meaner conoisient with Good Clinical Practices9. This Research Ethics Board hes reviewed and approved the chnicat trial protocol and informed concent form for the trial which is to be conducted t’ the qualifiedinvestigator named stove at the specified clinicel trial site. This approval and the views of this Research Ethics Board have been documented in writing.ho documentation included tsr the above-named project has bees reviewed by the USC CR85, and the research study, ae presanted in the docamantatbo, wastoued tobe acceptable unethical grounds for research invoIcing human oubjecto and wao spprsned by the USC ORES.Approval cIlIa ClinIcal Research Ethics Boardb cure rot1k Gall Beltward, ChaIr83


Citation Scheme:


Usage Statistics

Country Views Downloads
United States 5 0
India 3 0
China 3 0
Republic of Korea 2 0
Canada 2 0
Taiwan 1 0
City Views Downloads
Unknown 7 6
Beijing 3 0
Ashburn 2 0
Jinju 2 0
Taipei 1 0
Wilmington 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items