UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cardiac synchronized volumetric modulated arc therapy Poon, Justin 2018

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

Item Metadata


24-ubc_2018_september_poon_justin.pdf [ 5.74MB ]
JSON: 24-1.0371201.json
JSON-LD: 24-1.0371201-ld.json
RDF/XML (Pretty): 24-1.0371201-rdf.xml
RDF/JSON: 24-1.0371201-rdf.json
Turtle: 24-1.0371201-turtle.txt
N-Triples: 24-1.0371201-rdf-ntriples.txt
Original Record: 24-1.0371201-source.json
Full Text

Full Text

Cardiac Synchronized VolumetricModulated Arc TherapybyJustin PoonB.Sc. (Hons.), University of British Columbia, 2016A THESIS SUBMITTED IN PARTIAL FULFILLMENTOF THE REQUIREMENTS FOR THE DEGREE OFMaster of ScienceinTHE FACULTY OF GRADUATE AND POSTDOCTORALSTUDIES(Physics)The University of British Columbia(Vancouver)August 2018c© Justin Poon, 2018The following individuals certify that they have read, and recommend to the Fac-ulty of Graduate and Postdoctoral Studies for acceptance, the thesis entitled:Cardiac Synchronized Volumetric Modulated Arc Therapysubmitted by Justin Poon in partial fulfillment of the requirements for the degreeof Master of Science in Physics.Examining Committee:Dr. Steven ThomasSupervisorDr. Stefan ReinsbergSupervisory Committee MemberiiAbstractCardiac arrhythmias are abnormal heart rhythms that can adversely affect patients.The current treatment options include antiarrhythmic drug therapy and catheterablation, both of which are associated with issues regarding efficacy and potentialcomplications for complex arrhythmias.Cardiac radiosurgery, external beam therapy targeted at abnormal myocardialsubstrate, has the potential to be a non-invasive and efficient treatment option forarrhythmias. Intrafraction heart motion, however, must be accounted for to ensureaccurate dose delivery to the target region. Our technique aims to minimize thedose delivered to normal tissues by synchronizing beam delivery with a cardiacsignal, irradiating only during the quiescent intervals of the cardiac cycle (whenheart motion is minimal) and adjusting the beam delivery speed in response to achanging heart rate.Although real-time treatment plan adaptation is not possible with the currentlinear accelerator systems, we demonstrated the feasibility of our technique bysynchronizing beam delivery with sample ECG data. A conventional VMAT planwas split into 3 interleaved phases – each phase consists of alternating beam-onand beam-off arc segments, and the combination of all phases recreates the originalplan. Each phase was synchronized with a sample cardiac signal, using movementof the treatment couch to adjust the durations of individual arc segments. Thecardiac synchronized phases were created as XML beam files and were deliveredin developer mode of the Varian TrueBeam system. Analysis of the trajectory logsshowed that all beam-on segments were delivered during the quiescent intervals asplanned.Film dosimetry was used to verify the dose delivery accuracy of the cardiaciiisynchronized beams. A custom table was built and positioned over the treatmentcouch, allowing for a phantom placed on the table surface to remain stationary dur-ing beam delivery. Film was inserted in the phantom and irradiated using the origi-nal treatment plan, then a separate film was inserted and irradiated using the cardiacsynchronized phases. Analysis of the films in FilmQA Pro returned a gamma pass-ing rate of 99.4% (2%/2mm tolerance), indicating excellent agreement between thedose distributions of the original and cardiac synchronized beam deliveries.ivLay SummaryCurrent treatments for cardiac arrhythmias (abnormal heart rhythms) are often onlymodestly effective and may cause unwanted side effects or complications. Externalbeam radiation therapy targeted at the heart could be a safer and more efficientoption for arrhythmia treatment. The motion of the heart must be taken into accountto ensure accurate dose delivery. Our technique aims to minimize dose to normaltissue by turning the radiation beam on only during the relaxation phases of theheart cycle, when heart motion is minimal. Treatment machine parameters areadjusted as necessary to speed up or slow down the beam delivery in response to achanging heart rate. A treatment plan was successfully synchronized to a sampleheart signal and accurate dose delivery was demonstrated.vPrefaceThis thesis was completed by the author, Justin Poon, at the BC Cancer Vancouverand Fraser Valley Centres. The use of anonymized ECG data collected from a pre-vious project (REB #H13-01897) was approved by the UBC BC Cancer ResearchEthics Board (REB #H18-01231).viTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiiLay Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . viiList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xviiiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xx1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Cardiac Arrhythmia Mechanisms . . . . . . . . . . . . . . . . . . 21.1.1 ECG Signal . . . . . . . . . . . . . . . . . . . . . . . . . 41.2 Current Treatment of Cardiac Arrhythmias . . . . . . . . . . . . . 51.3 Cardiac Radiosurgery . . . . . . . . . . . . . . . . . . . . . . . . 51.3.1 Cardiac Synchronization Goals . . . . . . . . . . . . . . . 72 Creation of Modifiable Plans . . . . . . . . . . . . . . . . . . . . . . 92.1 Introduction to the TrueBeam Linear Accelerator . . . . . . . . . 92.2 Varian TrueBeam Developer Mode . . . . . . . . . . . . . . . . . 13vii2.2.1 Trajectory Model . . . . . . . . . . . . . . . . . . . . . . 142.2.2 XML Beam File . . . . . . . . . . . . . . . . . . . . . . 152.3 Trajectory Logs . . . . . . . . . . . . . . . . . . . . . . . . . . . 202.4 Creation of XML Beams . . . . . . . . . . . . . . . . . . . . . . 222.4.1 Machine Coordinate Conventions . . . . . . . . . . . . . 222.4.2 Converting IEC 61217 to Varian Scale . . . . . . . . . . . 222.5 XML Beam Delivery . . . . . . . . . . . . . . . . . . . . . . . . 243 Controlling the Beam Timing . . . . . . . . . . . . . . . . . . . . . . 273.1 Beam Delivery Time Calculation . . . . . . . . . . . . . . . . . . 283.2 Beam Timing Control Using Couch Movement . . . . . . . . . . 294 Creation of Interleaved Plans . . . . . . . . . . . . . . . . . . . . . . 354.1 Treatment Plan Interleaving . . . . . . . . . . . . . . . . . . . . . 354.1.1 Interleaver Code . . . . . . . . . . . . . . . . . . . . . . 374.2 Interleaving Results . . . . . . . . . . . . . . . . . . . . . . . . . 384.3 Arc Segment Timing Control . . . . . . . . . . . . . . . . . . . . 445 Cardiac Synchronization . . . . . . . . . . . . . . . . . . . . . . . . 475.1 Initial Attempt at Synchronization . . . . . . . . . . . . . . . . . 485.1.1 Beam Timing Calculation . . . . . . . . . . . . . . . . . 485.1.2 Beam Timing Correction . . . . . . . . . . . . . . . . . . 505.2 Low Dose Beam-Off Solution . . . . . . . . . . . . . . . . . . . 535.3 Cardiac Synchronization with Smooth Gantry Motion . . . . . . . 585.3.1 Dynamic Beam Timing . . . . . . . . . . . . . . . . . . . 585.3.2 Interleaving for Consistent Gantry Rotation Speed . . . . 605.3.3 Cardiac Synchronization . . . . . . . . . . . . . . . . . . 616 Treatment Accuracy Verification . . . . . . . . . . . . . . . . . . . . 656.1 Measurement Equipment and Treatment Planning . . . . . . . . . 656.2 Film Measurement . . . . . . . . . . . . . . . . . . . . . . . . . 686.2.1 Treatment Plan Dosimetry . . . . . . . . . . . . . . . . . 686.2.2 Calibration Curve . . . . . . . . . . . . . . . . . . . . . . 706.2.3 Film Analysis . . . . . . . . . . . . . . . . . . . . . . . . 71viii7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 767.1 Future Work . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79ixList of TablesTable 2.1 Trajectory log header structure showing available axis data. . . 21Table 3.1 Control of overall beam time using couch movement. . . . . . 30Table 7.1 Beam delivery time comparison. . . . . . . . . . . . . . . . . 76xList of FiguresFigure 1.1 Arrhythmia due to myocardial scarring. Top: Normal my-ocardium rapidly conducts an electrical wavefront, resulting insynchronous activation. Bottom: Scarred myocardium leads todisordered impulse propagation and dyssynchronous contraction. 3Figure 1.2 Each cardiac cycle in an electrocardiogram (ECG) signal has aP wave, QRS complex, and a T wave. A normal heart rhythmis shown on top, while an irregular rhythm is shown on thebottom. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4Figure 1.3 Synchronization of a linear accelerator beam to a cardiac cy-cle. The beam is turned on during the quiescent interval, whenheart motion is minimal, and turned off when the heart is con-tracting. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8Figure 2.1 The gantry angle coordinate system for Varian machines. Thegantry rotates about the isocentre slightly over 180◦ in eachdirection from the 0 position. . . . . . . . . . . . . . . . . . . 10Figure 2.2 The secondary collimator (upper and lower jaws) and tertiarycollimator (multileaf collimator (MLC)) as viewed when fac-ing the gantry. The upper jaws control the field size in the Y1and Y2 directions. The lower jaws control the field size in theX1 and X2 directions. The MLC consists of 2 opposing leafbanks. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11xiFigure 2.3 Diagram of a Varian TrueBeam linear accelerator. The gantryrotation axis and couch rotation axis are shown by the horizon-tal and vertical dotted lines, respectively. . . . . . . . . . . . . 13Figure 2.4 Gantry = F(MU) plot for a simple trajectory function with 2control points. The machine component will move linearlybetween each control point. . . . . . . . . . . . . . . . . . . 15Figure 2.5 Extensible Markup Language (XML) beam file for a simpleBeam ON plan, delivering 100 monitor unit (MU) while thegantry rotates from 0◦ to 180◦. The corresponding gantry tra-jectory plot is shown in Figure 2.4. . . . . . . . . . . . . . . . 16Figure 2.6 An example of an MLC element in an XML beam. The MLCelement is listed under a control point, and specifies the posi-tions of each leaf in banks B and A. . . . . . . . . . . . . . . 19Figure 2.7 An example of a tolerance table in an XML beam. The velocitytable is formatted identically. The tolerance and velocity tablesare inserted after the MLCModel element and before the Accselement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19Figure 2.8 Developer Mode screen showing option to collect trajectory log. 20Figure 2.9 Gantry angle conventions for IEC 61217 (left) and Varian scale(right). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23Figure 2.10 Collimator angle conventions for IEC 61217 (left) and Varianscale (right). . . . . . . . . . . . . . . . . . . . . . . . . . . 23Figure 2.11 Flowchart of converting a Digital Imaging and Communica-tions in Medicine (DICOM)-RT treatment plan to an XMLbeam using dicom2xml and delivering the beam in Devel-oper Mode. . . . . . . . . . . . . . . . . . . . . . . . . . . . 24Figure 2.12 Gantry rotation vs. time plot for DICOM and XML beam ver-sions of a volumetric modulated arc therapy (VMAT) plan. . . 25Figure 2.13 MU vs. time plot for DICOM and XML beam versions of aVMAT plan. . . . . . . . . . . . . . . . . . . . . . . . . . . 25Figure 3.1 Cumulative MU over time for beam deliveries with 0, 10, 15,and 20 cm couch translations. . . . . . . . . . . . . . . . . . 31xiiFigure 3.2 Control point progression over time for beam deliveries with0, 10, 15, and 20 cm couch translations. . . . . . . . . . . . . 32Figure 3.3 Control point vs. cumulative MU for beam deliveries with 0,10, 15, and 20 cm couch translations. . . . . . . . . . . . . . 32Figure 3.4 Gantry angle over time for beam deliveries with 0, 10, 15, and20 cm couch translations. . . . . . . . . . . . . . . . . . . . . 33Figure 3.5 Gantry angle vs. cumulative MU for beam deliveries with 0,10, 15, and 20 cm couch translations. . . . . . . . . . . . . . 33Figure 3.6 MLC leaf 90 position over time for beam deliveries with 0, 10,15, and 20 cm couch translations. . . . . . . . . . . . . . . . 34Figure 3.7 MLC leaf 90 position vs. cumulative MU for beam deliverieswith 0, 10, 15, and 20 cm couch translations. . . . . . . . . . 34Figure 4.1 The original plan is broken up into 400 ms portions, and eachportion is assigned to different phases in an alternating patternto create the interleaved plans. . . . . . . . . . . . . . . . . . 36Figure 4.2 The gantry angle vs. MU trajectory plot for an interleavedphase. Vertical segments are motion-only segments, wheregantry rotation continues but no MU is delivered. . . . . . . . 37Figure 4.3 Trajectory log results showing constant dose rate over time forthe original 2000 MU/min treatment plan (non-interleaved). . 39Figure 4.4 Trajectory log results for the first phase of a 2-phase inter-leaved plan with a dose rate of 2000 MU/min and 500 msbeam-on segments. . . . . . . . . . . . . . . . . . . . . . . . 40Figure 4.5 Trajectory log results of the interleaved phase showing the gantryrotation behaviour in relation to the dose rate. The gantry isforced to stop at every instance where the beam turns on or off. 40Figure 4.6 Trajectory log results showing constant dose rate over time forthe original treatment plan (non-interleaved). . . . . . . . . . 41Figure 4.7 Trajectory log results for the interleaved phase planned for 400ms beam-on and beam-off segments. . . . . . . . . . . . . . . 42xiiiFigure 4.8 Gantry angle and rotation speed over time for the 2-phase,800 MU/min, 400 ms beam-on interleaved plan. Instantaneousgantry stops were still causing an undesirable jerky motion. . . 43Figure 4.9 Planned cumulative MU and couch longitudinal motion for abeam delivery with 2 second beam-off intervals. The couchmoves 0.2 cm at 0.1 cm/s to force each beam-off segment tolast 2 seconds. . . . . . . . . . . . . . . . . . . . . . . . . . 44Figure 4.10 Trajectory log results showing dose rate and couch positionover time for an interleaved phase with 2 second beam-off in-tervals (2 phases, 800 MU/min, and 400 ms beam-on inter-vals). The average actual beam-off interval duration was 1.99seconds. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45Figure 4.11 Trajectory log results showing gantry position and speed overtime for an interleaved phase with 2 second beam-off intervals(2 phases, 800 MU/min, and 400 ms beam-on intervals). . . . 46Figure 5.1 Average of the previous 2 cardiac periods (yellow) is taken asthe predicted cardiac period. The beam-off duration (blue) iscalculated relative to the last R-peak and the predicted cardiacperiod, taking into account the time of the last beam-on. . . . 49Figure 5.2 Comparison between beam cycle times for an initial plan de-livery and a corrected plan delivery. The beam cycle times(beam-on and beam-off) were planned to be 2 seconds. . . . . 51Figure 5.3 Comparison of results between the uncorrected (top) and thecorrected (bottom) cardiac synchronized plan. In the correctedplan delivery, all beam-on segments are kept within the quies-cent intervals as planned. . . . . . . . . . . . . . . . . . . . . 52Figure 5.4 Cardiac synchronization is maintained as the heart rate increases25 bpm. The dose rate plot shown is the corrected plan delivery. 53xivFigure 5.5 Comparison between dose rate ramp times calculated by theTrueBeam control system for the old method (top) and the newlow dose method (bottom). For a maximum dose rate of 800MU/min, the dose rate ramp time was reduced from 140 to 40ms (or within 2 trajectory log snapshots). . . . . . . . . . . . 55Figure 5.6 Comparison between gantry rotation speed calculated by theTrueBeam control system for the old method (top) and the newlow dose method (bottom). The original plan was interleavedusing 8 phases. . . . . . . . . . . . . . . . . . . . . . . . . . 56Figure 5.7 No correction is needed for cardiac synchronization using thelow dose beam-off method. The reduced dose rate ramp timesallow for more accurate beam timing. . . . . . . . . . . . . . 57Figure 5.8 Beam-on segment timing for the low dose method matches theplanned timing (indicated by the green lines) without any cor-rection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58Figure 5.9 The previous cardiac period (yellow) is used to calculate thebeam-on duration (red). The beam-off duration (blue) is cal-culated relative to the last R-peak, taking into account the timeof the last beam-on. . . . . . . . . . . . . . . . . . . . . . . . 59Figure 5.10 Interleaving using 3 phases allows for smooth gantry motionwithin a cardiac period when the beam is turned on every heartbeat. Gantry speed is relatively consistent from beam-on tobeam-off during a phase delivery. The arc segment angulardisplacements are exaggerated for illustration purposes. . . . . 60Figure 5.11 Couch movement is increased or decreased in response to achanging heart rate, and dose rate is automatically adjusted.As the heart rate increases, couch translation is decreased toshorten the arc segment durations, and the dose rate increasesto maintain the planned dose delivery. . . . . . . . . . . . . . 62Figure 5.12 Beam delivery for a 3-phase interleaved plan successfully syn-chronized to an increasing heart rate. All beam-on intervalsare kept within the R-peaks. . . . . . . . . . . . . . . . . . . 63xvFigure 5.13 Comparison between gantry rotation speed calculated by theTrueBeam control system for a 3-phase interleaved plan us-ing the old method (top) and the new low dose method withdynamic beam-on times (bottom). The gantry rotation speedchanges in response to the changing heart rate, but the gantrymotion remains smooth between arc segments. . . . . . . . . 64Figure 6.1 A custom table was built for treatment accuracy verificationof the cardiac synchronized beam. The table allows for thetreatment couch to move freely underneath in the longitudinaldirection, while a phantom can be placed on the table surface. 66Figure 6.2 An SRS VMAT plan for the film measurement was created inthe Eclipse treatment planning system (TPS). The planningtarget volume (PTV) (red outline) and the planned gantry rota-tion are shown. . . . . . . . . . . . . . . . . . . . . . . . . . 67Figure 6.3 The isodoses for the treatment plan from the top-down view. . 67Figure 6.4 A QUASAR phantom with a wood insert and film cassette wasused for the treatment accuracy verification. A polystyrenefoam block was used to hold the phantom in the planned treat-ment position and a CT scan was taken for treatment planning. 68Figure 6.5 The phantom and custom table were used to take film mea-surements for both the original treatment plan and the cardiacsynchronized interleaved plan. . . . . . . . . . . . . . . . . . 69Figure 6.6 Solid water setup for the calibration curve film measurements.The films were placed at 100 cm source-to-axis distance (SAD)with 5 cm depth and 10 cm backscatter. A 10X flattening filterfree (FFF) beam energy and a 10 cm x 10 cm field size wereused. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70Figure 6.7 Films used for the calibration curve. From left to right: 650,433.3, 288.9, 192.6, 128.4, 85.6, 57.1, and 0 (unexposed) cGy. 71Figure 6.8 Film scan for the original treatment plan delivery. A 650 cGyfilm and an unexposed film were used as reference films. . . . 72xviFigure 6.9 Film scan for the cardiac synchronized interleaved plan de-livery. A 650 cGy film and an unexposed film were used asreference films. . . . . . . . . . . . . . . . . . . . . . . . . . 72Figure 6.10 FilmQA Pro dose isolines comparing the original treatmentplan and the cardiac synchronized interleaved plan. . . . . . . 74Figure 6.11 FilmQA Pro dose profile comparing the original treatment plan(light red) and the cardiac synchronized interleaved plan (darkred). The line profile was drawn horizontally across the centreof the irradiated area. . . . . . . . . . . . . . . . . . . . . . . 75xviiGlossaryAF atrial fibrillationAV atrioventricularCT computed tomographyDD dose differenceDIBH deep inspiration breath holdDICOM Digital Imaging and Communications in MedicineDTA distance-to-agreementECG electrocardiogramFFF flattening filter freeFWHM full width at half maximumIEC International Electrotechnical CommissionMLC multileaf collimatorMU monitor unitPTV planning target volumeRF radiofrequencyRP radiotherapy planxviiiSA sinoatrialSABR stereotactic ablative radiotherapySAD source-to-axis distanceSBRT stereotactic body radiotherapySRS stereotactic radiosurgerySSD source-to-surface distanceTPS treatment planning systemVMAT volumetric modulated arc therapyVT ventricular tachycardiaXML Extensible Markup LanguagexixAcknowledgmentsI would like to acknowledge all the staff at the BC Cancer Agency Vancouver andSurrey centres, who were always approachable and happy to help. A special thankyou to Cheryl Duzenli for her continuous support and for giving me the opportu-nity to spend my graduate studies with a wonderful group in the medical physicsdepartment. Thanks to Don Ta, Joel Beaudry, and Claudia Mendez for their thor-ough training in linac operation, allowing me to be comfortable experimentingwith million-dollar machines. I would also like to thank Tania Karan, Marie-LaureCamborde, and Stanley Szpala for their help with film measurements and analysis.Thanks to Gina Badragan for her teaching and guidance at the Fraser Valley Cen-tre, giving me the confidence to independently carry out late-night measurementsfor my project. To those involved directly with the project – Kirpal Kohli, DevinSchellenberg, Marc Deyell, and Stefan Reinsberg – I feel incredibly privileged tobe working with you and am grateful for all the advice and motivation you havegiven me. To Steven Thomas, thank you for your mentorship, for being a fantasticsupervisor, and for giving me the privilege to work on an exciting project that hasled to amazing opportunities and will lead to many more in the future. Lastly, Iwould like to give a huge ‘thank you’ to my family for their unconditional supportand endless encouragement.xxChapter 1IntroductionCardiac arrhythmias are abnormal heart rhythms or heart rates that can adverselyaffect patients. An arrhythmia may take the form of an irregular heart rhythm (flut-ter or fibrillation), an abnormally slow heart beat (bradycardia), or a rapid heart beat(tachycardia). While some arrhythmias have minimal impact on a patient’s health,others may indicate structural heart disease or increase the risk of stroke or sud-den cardiac arrest.1,2 Patients may experience heart palpitations (skipped or extraheartbeats) in milder cases, while more serious symptoms include lightheadedness,fainting, shortness of breath, and chest pain.3 Recurrent or sustained arrhythmiasare associated with reduced quality of life, increased risk of heart failure, and re-duced survival.4,5The incidence for most arrhythmias increases with age, although certain con-genital arrhythmias may arise in children.6 In the United States, atrial fibrilla-tion (AF) is the second-most common cardiovascular condition7 and was estimatedto have affected 3 to 6 million Americans in 2010.8 AF was estimated to affect 1%to 1.5% of the population in the developed world – with the aging population, thisnumber is expected to triple by 2050.9 Cardiovascular disease is also the leadingcause of global mortality (17 million deaths per year or 30% of global mortality) –it is estimated that about half of all cardiovascular-related deaths are sudden cardiacdeaths, 80% of which are caused by ventricular tachyarrhythmias.1011.1 Cardiac Arrhythmia MechanismsTo understand the mechanisms behind cardiac arrhythmias, it is important to befamiliar with the electrical system of the heart. The heart is responsible for pump-ing blood through the body’s circulatory system and contains 4 chambers: the leftatrium, left ventricle, right atrium, and right ventricle. Deoxygenated blood fromthe body enters the right atrium, is passed to the right ventricle, and is pumpedthrough the pulmonary artery to the lungs for reoxygenation. The oxygenatedblood returns to the left atrium through the pulmonary vein, enters the left ven-tricle, and is pumped through the aorta to the rest of the body.11The beating of the heart is the result of rhythmic contractions of the atria andventricles, which are activated by electrical signals generated by the cardiac con-duction system.12 Cardiac myocyte (heart cell) contraction is initiated when thecell is depolarized by an electrical impulse or action potential. The cardiac im-pulse originates from pacemaker cells within the sinoatrial (SA) node and spreadsthrough the atrium, propagating through the myocardium (heart muscle) by way ofintermyocyte connections called gap junctions. The impulse is conducted throughinternodal pathways to the atrioventricular (AV) node, and then spreads through theventricular myocardium. Arrhythmias may arise from irregularities in the cardiacconduction system, such as abnormal impulse formation (altered automaticity ortriggered activity), abnormal impulse transmission (reentry), or a combination ofboth.13Automaticity is the ability of myocytes to spontaneously depolarize in the ab-sence of an external electrical stimulation (as in the SA node). Enhanced normalautomaticity refers to accelerated action potential generation from normal pace-maker cells. Abnormal automaticity occurs when cells that do not normally initiateimpulses acquire automaticity, leading to extra heart beats.Triggered activity refers to impulse initiation caused by afterdepolarizations,which are depolarizing oscillations that occur following an action potential. Af-terdepolarization may occur during the normal repolarization phase (early after-depolarization) or after the repolarization phase (late afterdepolarization). A newimpulse is triggered when an afterdepolarization reaches the threshold potential,causing arrhythmia.2Reentry is where a propagating impulse fails to die out after normal heart con-traction and continues to reexcite the heart. It is the mechanism responsible forthe majority of clinically important arrhythmias.14 Reentry can occur if the car-diac impulse encounters a nonconductive blocked area, loops around the blockedzone, and reenters the original excitation site. The impulse repetitively propagatesthrough this “circuit” and continuously excites heart tissue, leading to tachycar-dia. The tachycardia should terminate when a limb of the pathway is removed orblocked.15Scar tissueFigure 1.1: Arrhythmia due to myocardial scarring. Top: Normal my-ocardium rapidly conducts an electrical wavefront, resulting in syn-chronous activation. Bottom: Scarred myocardium leads to disorderedimpulse propagation and dyssynchronous contraction.The anatomic substrates that cause arrhythmias may be congenital or arise frompathologic changes to heart tissue.16 Congenital accessory pathways are abnormalmyocardium that bypass the normal conduction system and can create a reentrantcircuit. Scarred myocardium can also act as conduction blocks or constraints for3the reentrant circuit (Figure 1.1). Post-myocardial infarction (heart attack) -relatedscarring is a common cause of ventricular tachycardia (VT). Arrhythmias mayalso arise from unavoidable scars left behind after cardiac surgery.17 Additionally,aging and progression of heart disease can lead to increased fibrosis (hardening andscarring of tissue) which can promote arrhythmias.181.1.1 ECG SignalAn electrocardiogram (ECG) is a device that uses electrodes placed on the skin tomeasure the heart’s electrical activity. An atrial depolarization wave (P wave), aventricular depolarization wave (QRS complex), and a ventricular repolarizationwave (T wave) are recorded for each cardiac cycle.19 The top of the QRS complexis referred to as the R-peak. Figure 1.2 shows ECG signals for a normal heartrhythm and an irregular heart rhythm.P waveQRS complexT waveNormal rhythmIrregular rhythmFigure 1.2: Each cardiac cycle in an ECG signal has a P wave, QRS complex,and a T wave. A normal heart rhythm is shown on top, while an irregularrhythm is shown on the bottom.41.2 Current Treatment of Cardiac ArrhythmiasCurrent cardiac arrhythmia treatments include antiarrhythmic drug therapy or minimally-invasive catheter ablation. Both methods, however, are associated with issues re-garding efficacy and potential complications.Antiarrhythmic drugs, in general, alter the cardiac myocyte ion channels tochange the shape of the action potential.20 By binding to the channels that controlthe flow of ions across cardiac cell membranes, these drugs modify cardiac tissueconduction velocity, refractoriness (minimum time between excitations), and auto-maticity. Unfortunately, antiarrhythmic drugs are often weakly effective and canresult in cardiovascular and non-cardiovascular toxicities.21In catheter ablation, a flexible tube (catheter) is threaded through the patient’sblood vessels towards the target area of the heart.22 An electrode at the catheter tipemits radiofrequency (RF) energy to heat the nearby tissue, leading to tissue necro-sis and disruption of the arrhythmogenic substrate.17 Success rates and complica-tion risks for catheter ablation are dependent on the targeted substrate – VT and AFhave the lowest reported success rates.23,24 Inadequate heating and scar formationat the ablation target can lead to the return of electrical activity in arrhythmogenictissue. Larger targets are also associated with higher recurrence rates due to greaterrecovery of electrical conduction. Target location may also make catheter ablationunsuitable – critical structures such as the SA/AV nodes, esophagus, or phrenicnerve may be in close proximity, or the target may be inaccessible when locateddeep beneath the myocardial surface.23 There is a major clinical need for a treat-ment option that is more consistent and efficient.1.3 Cardiac RadiosurgeryCardiac radiosurgery could potentially serve as a completely non-invasive treat-ment option for cardiac arrhythmias. Unlike catheter ablation, which uses RF en-ergy to heat tissue, radiotherapy uses photons or electrons to affect the targetedtissue. A linear accelerator is used to focus a beam of ionizing radiation to de-liver dose to the target region (more information about the linear accelerator andits components will be provided in Section 2.1). The absorbed dose is a measureof the radiation energy deposited per unit mass in J/kg or Gray (Gy).5Modern techniques such as volumetric modulated arc therapy (VMAT) allowfor precise dose delivery to a target (usually a cancerous tumour) while minimizingdose delivered to adjacent normal tissues.25 During VMAT delivery, the radiationdose is continuously delivered while the treatment machine rotates. The multileafcollimator (MLC) leaves, which are used to shape the beam, are also moving con-tinuously during delivery, producing a highly conformal dose distribution.The development of the stereotactic radiosurgery (SRS) technique allowed forablation of brain tumours with submillimeter accuracy.26 The application of ra-diosurgery outside of the brain is also referred to as stereotactic body radiother-apy (SBRT) or stereotactic ablative radiotherapy (SABR), and can be used to treattumours in areas such as the lung, liver, and spine.27 In contrast to conventionalradiotherapy, where treatment is delivered in daily fractions over several weeks,SBRT treatments involve high doses to a small lesion (less than 5 cm diameter) in1 to 5 fractions.27SBRT has been safely used to treat a tumour located in the heart.28 The appli-cation of SBRT to the treatment of cardiac arrhythmias has recently gained interest.CyberHeart Inc. has developed a cardiac radiosurgery system using the CyberKnife(Accuray Inc.) robotic radiosurgery device.29 Cuculich et al from Washington Uni-versity have also used SBRT to treat 5 patients with VT.30 Electrocardiographicimaging, a non-invasive method for mapping cardiac electrical activity,31 was usedto identify and localize myocardial scar tissue causing VT. The arrhythmogenicscar regions were ablated using SBRT in a single fraction of 25 Gy. A reduction inVT episodes occurred in all 5 patients.30Cardiac radiosurgery also has the potential to provide higher cost-effectivenessand improved patient outcomes compared to current treatment options.7 Costs as-sociated with invasive procedures such as recovery time in the hospital would beeliminated. There would also be a reduction in re-hospitalizations occurring due tovascular complications with the catheter ablation procedure. Catheter ablation forthe treatment of AF has a reported rate of recurrence between 50% to 80% for 1 to3 years after the initial procedure.7 Repeat ablation procedures are associated withhigher complication risks and increased total costs.61.3.1 Cardiac Synchronization GoalsCardiac radiosurgery appears to be a promising treatment option for arrhythmias,but intrafraction heart motion is a major complicating factor for precise beam deliv-ery. Motion associated with both the respiratory and cardiac cycles must be takeninto account to ensure accurate dose delivery to the planning target volume (PTV).While patient breathing is responsible for a larger portion of target motion in thechest region, it is more easily controllable using techniques such as deep inspira-tion breath hold (DIBH).32 Cardiac motion, on the other hand, presents a moredifficult problem since the heart rate cannot be voluntarily controlled.The period from the beginning of one heartbeat to the beginning of the nextis defined as the cardiac cycle, consisting of a systolic (contraction) phase and adiastolic (relaxation) phase. Heart motion is the greatest during the systolic phaseof the cardiac cycle. In a study by Wang et al 33 , heartbeat-related motion in 20patients was measured by monitoring displacement of the left anterior descendingartery during DIBH. Average maximum displacement was 7.1 mm in the left-rightdirection and 8.2 mm in the anterior-posterior direction. Motion is minimal duringthe diastolic phase, when the heart is relaxed. Cardiac synchronized radiotherapyaims to minimize dose to normal tissue by irradiating only during these relax-ation phases or quiescent intervals. Previous cardiac radiotherapy applications, incomparison, simply use an expanded PTV to account for heart motion, deliveringunnecessary dose to areas surrounding the arrhythmogenic target region.29,30For a typical heart rate of 75 beats per minute (bpm), the cardiac cycle is 0.8seconds long.60s/min75beats/min= 0.8s/beat (1.1)During this period, the atria are in systole for 0.1 seconds and in diastole for 0.7 sec-onds. Ventricular systole and diastole last for 0.3 and 0.5 seconds, respectively.34Joint diastole – when atrial and ventricular diastole overlap and all four chambersof the heart are at rest – lasts for about 0.4 seconds. The goal of cardiac synchro-nization is to turn the linear accelerator beam on during this window of time, whenheart motion is at a minimum, and off during the rest of the cardiac period (Figure1.3). The technique should allow for the linear accelerator to alternate betweenshort beam-on and beam-off intervals while the gantry continuously rotates, essen-7AtrialsystoleAtrialsystoleAtrialdiastoleVentriculardiastoleVentricularsystoleVentriculardiastoleBeam off Beam offBeam onFigure 1.3: Synchronization of a linear accelerator beam to a cardiac cycle.The beam is turned on during the quiescent interval, when heart motionis minimal, and turned off when the heart is contracting.tially delivering a cardiac synchronized VMAT plan. The linear accelerator motionmust be kept as smooth as possible, as sudden and excessive acceleration wouldresult in less accurate delivery due to the large momentum of the rotating gantry.25The technique must also demonstrate beam delivery adaptation to a changingheart rate. All machine components should be sped up or slowed down simulta-neously in response to an increasing or decreasing heart rate, maintaining treat-ment plan accuracy and synchronization of the beam-on intervals with the quies-cent phases. For example, an increase in heart rate should lead to an increasedfrequency of beam-on intervals. Gantry rotation speed, MLC leaf speeds, and doserate must be increased in order for the prescribed dose to be delivered as planned,while having the beam continue to turn on at the desired times.8Chapter 2Creation of Modifiable PlansAll development work for the cardiac synchronization technique was done on Var-ian TrueBeam linear accelerators. Although real-time plan adaptation is not pos-sible with the current TrueBeam system, we could still demonstrate the conceptof cardiac synchronization by manipulating the beam delivery of a conventionalVMAT plan. Manipulation of beam timing in the precise fashion required by ourtechnique is not achievable using the Eclipse treatment planning system (TPS). Amethod that allowed for experimentation and easy modification of machine param-eters was required, so the TrueBeam system’s Developer Mode was used.2.1 Introduction to the TrueBeam Linear AcceleratorFirst, a brief introduction to the Varian TrueBeam system is provided. The True-Beam linear accelerator is a treatment machine used for external beam radiationtherapy, where high energy beams of photons or electrons are directed at a targetregion in the patient’s body.The gantry is the main C-shaped portion which contains the electron gun, theaccelerator structure, and the treatment head. The electron gun creates a streamof electrons that are accelerated by the action of radio-frequency waves along thewave guide towards the treatment head. For photon therapy, the electrons are di-rected by bending magnets towards a tungsten target. Bremsstrahlung interactionsin the tungsten target result in the production of photons, and the photon beam90°90°180°270°315°225°45°135°Figure 2.1: The gantry angle coordinate system for Varian machines. Thegantry rotates about the isocentre slightly over 180◦ in each directionfrom the 0 position.exits the treatment head in the direction of the patient. The gantry rotates about apoint in space called the isocentre, which is located on the beam axis 100 cm fromthe radiation source. The gantry can rotate just over 180◦ in each direction fromthe 0 position (see Figure 2.1). Ideally, the mechanical isocentre is identical to theradiation isocentre – the intersection point of radiation beams from different gantryangles.The treatment head also contains primary, secondary, and tertiary collimatorsthat modify the beam shape. The primary collimator is located closest to the X-10Y1X1X2Y2Bank BBank AUpper jawsLower jawsMLCFigure 2.2: The secondary collimator (upper and lower jaws) and tertiary col-limator (MLC) as viewed when facing the gantry. The upper jaws con-trol the field size in the Y1 and Y2 directions. The lower jaws controlthe field size in the X1 and X2 directions. The MLC consists of 2 op-posing leaf banks.ray target and defines the maximum circular field size. The secondary collimatorconsists of upper and lower jaws that shape the beam in the Y and X directions,forming a rectangular field size. The tertiary collimator is the multileaf collima-tor (MLC), which is made up of tungsten leaves that are used to further shapethe beam field. Varian TrueBeam linear accelerators contain 120 leaf MLCs (60pairs of opposing leaves). The leaf widths may either be 5 mm (Millennium MLCmodel) or 2.5 mm (High Definition MLC model). The collimator or treatment headcan also rotate about the beam axis 180◦ in each direction. Figure 2.2 shows thecollimator arrangement and naming system.A flattening filter is located between the primary and secondary collimators.11This filter is made of high Z materials, and its conical shape works to flatten theforward-peaked Bremsstrahlung spectrum of megavoltage photons. The presenceof a flattening filter creates a flat dose profile at reference depth, but also resultsin a substantially reduced photon beam dose rate.35 The filter can also be movedout of the beam path to operate the linear accelerator in flattening filter free (FFF)mode, allowing for higher dose rates to be achieved.Below the flattening filter are the monitor chambers – two ionization chambersthat measure the machine output in monitor units (MUs). The monitor chambersare calibrated to read 100 MU when an absorbed dose of 1 Gy is delivered to apoint. The location of this calibration point varies between treatment centres, butis most commonly defined using one of two setups:1. 100 cm source-to-surface distance (SSD), 10 cm x 10 cm field size.The surface of a water phantom is at the isocentre (100 cm SSD), and thecalibration point is at the depth of maximum dose.2. 100 cm source-to-axis distance (SAD), 10 cm x 10 cm field size.The water phantom is positioned such that the calibration point is at theisocentre (100 cm SAD).The treatment couch, on which the patient lies, can be moved in the longitudi-nal (parallel to the gantry rotation axis) and lateral directions (perpendicular to thegantry rotation axis). The couch is also height-adjustable and can be rotated aboutthe couch/beam axis.12GantryElectron gunAccelerating waveguide Bending magnetX-ray targetPrimary collimatorFlattening filterIon chamberUpper jawsLower jawsMultileaf collimatorIsocentreCouchFigure 2.3: Diagram of a Varian TrueBeam linear accelerator. The gantryrotation axis and couch rotation axis are shown by the horizontal andvertical dotted lines, respectively.2.2 Varian TrueBeam Developer ModeThe TrueBeam Developer Mode is functionally identical to the clinical modes usedduring regular treatment delivery, but enables access to additional advanced con-trol features. The defining feature of Developer Mode is the unique format ofthe treatment plans, which allows for full control of the TrueBeam system’s ca-pabilities. Plans loaded in Treatment and Service Mode are typically in DICOM-RT format, which is an extension of the Digital Imaging and Communicationsin Medicine (DICOM) standard used in diagnostic imaging. The DICOM-RTformat was developed for use in radiation therapy and supports the transfer ofradiotherapy-related data, including prescribed dose and all treatment plan param-eters. In Developer Mode, however, plans are formatted in Extensible Markup13Language (XML) and are therefore called XML beams. Each XML beam containsa list of sequential control points that defines the machine’s trajectory throughoutthe plan.2.2.1 Trajectory ModelThe trajectory model is a concept that is used throughout the TrueBeam controlsystem architecture, where each beam is described as a relationship between MUand position. The trajectory function describes the position of a mechanical axisas a function of cumulative MU (the total MU delivered up to a certain point).Axis Position = F(MU)The collection of trajectory functions for all axes is referred to as the machinetrajectory.Each trajectory function is defined by a series of discrete points called controlpoints. In other words, each control point specifies how the machine componentshould be positioned when a certain number of MU has been delivered. The linearpath between two consecutive control points is referred to as a segment, and theentire segmented trajectory defines the axis motion during beam delivery.For example, a simple trajectory function for gantry angle may have 2 controlpoints (Figure 2.4):Control Point 0: Gantry Angle = 0◦ and MU = 0Control Point 1: Gantry Angle = 180◦ and MU = 100Machine motion remains linear during the arc segment between the 2 control points.The gantry angle will be at 45◦ when the cumulative MU reaches 25 and 90◦ whenthe cumulative MU reaches 50.If no motion occurs between 2 control points but radiation is delivered, thatsegment is referred to as a beam-only or static segment. A static segment wouldbe depicted on a trajectory plot as a horizontal line. If there is motion between2 control points but no change in MU, that segment is called a motion-only orbeamless segment, and is represented as a vertical line on the trajectory plot.140 20 40 60 80 100MU050100150200Gantry Angle (deg)Control Point 0Gantry Angle = 0°MU = 0Control Point 1Gantry Angle = 180°MU = 100Gantry Angle = 90°MU = 50Figure 2.4: Gantry=F(MU) plot for a simple trajectory function with 2 con-trol points. The machine component will move linearly between eachcontrol point.2.2.2 XML Beam FileThe XML beam file contains a list of sequential control points, where each controlpoint specifies instantaneous machine parameters. This allows the user to essen-tially program the machine trajectory by defining control points and synchronizingaxes positions with cumulative MU.XML files are used to store and exchange information in text format. An XMLfile consists of information packets called ”elements” that are organized into a hier-archical tree structure. Attributes define element properties, and each elements canhave a number of attribute/value pairs. An XML Schema describes the structure ofan XML document, specifying rules such as:• Legal elements and attributes15• Default and fixed values for elements and attributes• Mandatory and optional child elements• Order of child elements• Valid data typesXML beams must follow a specific ”Varian TrueBeam XML Schema”, which spec-ifies the valid syntax and content expected in Developer Mode.1 <VarianResearchBeam SchemaVersion="1.0">2 <SetBeam>3 <Id>1</Id>4 <MLCModel>NDS120HD</MLCModel>5 <Accs></Accs>6 <ControlPoints>7 <Cp>8 <SubBeam>9 <Seq>0</Seq>10 <Name>Beam ON</Name>11 </SubBeam>12 <Energy>6x</Energy>13 <Mu>0</Mu>14 <DRate>400</DRate>15 <GantryRtn>180.0</GantryRtn>16 </Cp>17 <Cp>18 <Mu>100</Mu>19 <GantryRtn>0.0</GantryRtn>20 </Cp>21 </ControlPoints>22 </SetBeam>23 </VarianResearchBeam>Figure 2.5: XML beam file for a simple Beam ON plan, delivering 100 MUwhile the gantry rotates from 0◦ to 180◦. The corresponding gantrytrajectory plot is shown in Figure 2.4.Figure 2.5 shows an XML beam file for the simple example described in Sec-tion 2.2.1. The general format for an XML beam begins with the machine details16and initial parameters followed by the list of control points. The mandatory ele-ments that must be present in all XML beams are described below:<VarianResearchBeam> is always the first element, and the SchemaVersionattribute ensures version compatibility with the target TrueBeam machine.The <SetBeam> element is the overall container for the XML beam elements.<Id> allows the user to assign an integer identifier for the beam.<MLCModel> defines the linear accelerator MLC model that is expected.<Accs> lists any accessories that must be mounted before beam delivery. If leftempty, the system will not care if any accessories are mounted or not.The <ControlPoints> element contains the list of control points that define theoverall machine trajectory. A minimum of 2 control points (one segment) is re-quired. Each <Cp> element defines one control point and contains the associatedparameters. The first control point contains a number of mandatory elements:The <SubBeam> element contains a sequence identifier and a string name – anoverall plan can be made up of multiple beam fields, which are defined by individ-ual XML files in Developer Mode.The <Energy> element specifies the beam energy, where the letter indicates theparticle type (x for photons and e for electrons). This element can only appear inthe first control point, as the beam energy cannot be changed partway through plandelivery.<Mu> is the cumulative MU, which must be 0 in the first control point.<DRate> defines the dose rate in MU per minute. In a dynamic beam where themachine is in motion while the beam is on, this parameter acts as a dose rate ceilingthat will never be exceeded.The first control point may include any number of mechanical axis positions afterthe mandatory elements have been defined. For the example in Figure 2.5, the17initial gantry angle is specified in the <GantryRtn> element. Axis positions thatare not included in the first control point are not checked by the control system– these machine components are allowed to be at any position. If any machineparameters such as MLC model, mounted accessories, or initial axis positions donot match the XML specifications at the time of beam loading, the system willprevent beam-on through an interlock.Control points after the first only need to include elements that involve a changein position. Axes that are not included are assumed to be stationary, remaining atthe position defined in some previous control point. In the Figure 2.5 example,the second control point only specifies the final cumulative MU and gantry angle.This simple 2-control point XML beam will tell the TrueBeam system to rotate thegantry from 0◦ to 180◦ while delivering 100 MU at a dose rate of 400 MU/min.(Note: The gantry angles used in XML beams follow the Varian machine scalerather than the International Electrotechnical Commission (IEC) standard. SeeSection 2.4.1.)The <Mlc> control point element is a special case – if any individual MLCleaf changes position, then the full array of leaf positions must be included. The<Mlc> element contains separate subelements for leaf banks B and A, in whichthe position of each individual leaf is specified.A tolerance table and velocity limit table can also be included before the con-trol point list as optional elements. The <TolTable> element is used to specifyposition tolerances for specific machine axes. The actual and planned machineaxis positions must be within tolerance before the beam can be turned on. The<VelTable> element allows the user to specify the maximum speed at which amachine axis can move. The default velocity limits are used if no velocity limitsare given.Units: Positions in XML beam files are given in centimetres for translational axesor degrees for rotational axes.181 <Mlc>2 <ID>1</ID>3 <B>0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.050.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.050.05 0.05 0.031 0.219 0.375 1.406 1.656 1.906 2.906-0.844 1.156 1.0 0.969 0.469 0.281 0.156 0.05 0.050.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.050.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.05</B>4 <A>-0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05-0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05-0.05 -0.05 -0.05 -0.05 -0.05 0.02 -0.168 -0.031 1.5-0.375 -0.625 -0.406 2.906 2.656 0.25 1.625 1.0940.656 0.156 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05-0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05-0.05 -0.05 -0.05 -0.05 -0.05 -0.05 -0.05</A>5 </Mlc>Figure 2.6: An example of an MLC element in an XML beam. The MLCelement is listed under a control point, and specifies the positions ofeach leaf in banks B and A.1 <TolTable>2 <GantryRtn>1</GantryRtn>3 <CollRtn>1</CollRtn>4 <CouchVrt>2.0</CouchVrt>5 <CouchLat>10.0</CouchLat>6 <CouchLng>10.0</CouchLng>7 <CouchRtn>1</CouchRtn>8 <Y12>0.2</Y12>9 <X12>0.2</X12>10 </TolTable>Figure 2.7: An example of a tolerance table in an XML beam. The veloc-ity table is formatted identically. The tolerance and velocity tables areinserted after the MLCModel element and before the Accs element.192.3 Trajectory LogsTrueBeam machines have the ability to output a trajectory log file following beamdelivery. Trajectory logs are binary files that contain position information for everymachine axis throughout the beam delivery in 20 millisecond (ms) intervals. Everysample point includes both expected and actual machine axis positions. The ex-pected positions are calculated from the given control points, and are determinedat the time of loading the beam and pressing the prepare button on the controlconsole. The TrueBeam system also records actual axis positions during delivery.Table 2.1 shows the trajectory log header structure and the available axis data.The control point axis data are floats, indicating the segment progress at each sam-ple point. For example, a control point value of 1.5 means that sample point ishalfway between the first and second control point.A Python module called loganalyzer was created to aid in trajectory loganalysis. The trajectory log binary file is read with the help of the Python modulepylinac created by James Kerns,36 and the axis position and speed informationis analyzed using our loganalyzer module. Our program is also capable ofcreating position or speed versus time or MU plots for every machine axis, whichallowed us to easily visualize and compare trajectory log results.Figure 2.8: Developer Mode screen showing option to collect trajectory log.20Table 2.1: Trajectory log header structure showing available axis data.Data Description Size TypeSignature ’VOSTL’ 16 bytes stringVersion ’3.0’ 16 bytes stringHeader size (fixed at 1024) 4 bytes integerSampling interval in milliseconds 4 bytes integerNumber of axes sampled 4 bytes integerAxis enumeration Number of axes * 4 bytes integer arrayColl Rtn – 0Gantry Rtn – 1Y1 – 2Y2 – 3X1 – 4X2 – 5Couch Vrt – 6Couch Lng – 7Couch Rtn – 9MU – 40Beam Hold – 41Control Point – 42MLC – 50Samples per axis Number of axes * 4 bytes integer arrayAxis Scale 4 bytes integer1 – Machine Scale2 – Modified IEC 61217Number of subbeams 4 bytes integerIs Truncated? 4 bytes integerNumber of snapshots 4 bytes integerMLC model 4 bytes integer2 – NDS 1203 – NDS 120 HD212.4 Creation of XML BeamsIn order to simplify the process of creating complex XML beams, we created aPython program called dicom2xml to convert regular DICOM plans to XML filescompatible with Developer Mode. Our first goal was to simply take an inputtedDICOM radiotherapy plan (RP) file and output the equivalent plan in XML format,without modifying any plan parameters, and ensure that both versions result inidentical plan delivery.The dicom2xmlmodule contains a class called XmlBeam, whose main methodwrite parses an inputted DICOM file and creates the equivalent XML file. All re-quired parameters are read from the DICOM file attributes using the dicom Pythonmodule, and the built-in module ElementTree is used to create the XML beamtree structure described in Section 2.2.2.The control points in the DICOM plan’s ControlPointSequence are parsedone by one, and the parameters are converted to the correct scale and format beforebeing written to the equivalent XML beam elements. The first control point alwayscontains the most information, including the beam energy and initial machine axispositions. The rest of the control points only include parameters that change – fora typical VMAT plan, this includes cumulative MU, gantry angle, and MLC leafpositions.2.4.1 Machine Coordinate ConventionsIt is important to use the correct coordinate convention when working with bothDICOM and XML format treatment plans. DICOM-RT plans follow the Inter-national Electrotechnical Commission (IEC) 61217 coordinate convention, whileXML beams and trajectory logs use the Varian machine scale. Information parsedfrom DICOM files must be converted to Varian scale before being written to XMLbeam files, so that the equivalent plan may be delivered.2.4.2 Converting IEC 61217 to Varian ScaleTranslational axis positions are given in millimetres (mm) for the IEC 61217 con-vention and in centimetres (cm) for Varian scale. For IEC scale, the gantry anglebegins at 0◦ (treatment head at the top) and increases in the clockwise direction220°90°180°270°180°90°0°270°IEC Scale Varian ScaleFigure 2.9: Gantry angle conventions for IEC 61217 (left) and Varian scale(right).0°90°270° 90°270°180°IEC Scale Varian ScaleFigure 2.10: Collimator angle conventions for IEC 61217 (left) and Varianscale (right).as the viewer faces the gantry. Likewise, the initial collimator angle is 0◦ and in-creases in the clockwise direction (looking up through the collimator and towardsthe source). In Varian scale, the zero angle is offset from the IEC convention by180◦ and increases in the counter-clockwise direction.The dicom2xml program handles translational axis position conversion by23simply dividing the IEC scale value by 10 and adding an offset for certain axes.Angles are converted from IEC to Varian scale using the following equation:(180−a) mod 360 (2.1)where a is the angle in IEC scale and mod is the modulo operator.2.5 XML Beam DeliveryThe functionality of the dicom2xml program was tested by delivering DICOMand XML beam versions of the same treatment plan and analyzing the trajectorylogs. A simple VMAT plan with a 180 degree arc was created using the EclipseTPS. The DICOM file was then converted to an XML beam using the dicom2xmlprogram. The original DICOM plan was first delivered on a TrueBeam linear ac-celerator and the trajectory log was recorded. The XML beam was delivered inDeveloper Mode, then the trajectory logs were analyzed to check if the DICOMand XML versions were indeed equivalent.dicom2xmlDICOM XML BeamTrueBeamDeveloper ModeDeliver beamTrajectory logsFigure 2.11: Flowchart of converting a DICOM-RT treatment plan to anXML beam using dicom2xml and delivering the beam in DeveloperMode.24Figure 2.12: Gantry rotation vs. time plot for DICOM and XML beam ver-sions of a VMAT plan.Figure 2.13: MU vs. time plot for DICOM and XML beam versions of aVMAT plan.25Figures 2.12 and 2.13 show the gantry angle and MU results from the DICOMand XML beam trajectory logs. The beam delivery was identical for both versionsof the treatment plan. Our tool for DICOM plan to XML beam conversion wassuccessful, allowing us to begin experimenting with plan modification for cardiacsynchronization.26Chapter 3Controlling the Beam TimingSynchronization of linear accelerator beam delivery to a cardiac signal requiresfine control over the beam timing. The beam delivery speed must be adaptable toa varying heart rate. For example, the dose rate, gantry rotation speed, and MLCleaf speeds would need to be increased in response to a faster heart beat so that theplanned radiation is delivered within the shorter quiescent intervals. Similarly, aslower heart beat would require a decrease in the above mentioned parameters inorder to spread out the beam delivery over the longer quiescent intervals.Cardiac synchronization of a VMAT plan requires that the dose delivered dur-ing each arc segment remains as planned, regardless of any modification of othermachine parameters. The rate of dose delivery with respect to gantry angle is re-lated to dose rate and gantry rotation speed by the following equation:dMUdθ=dMUdt/dθdt(3.1)In the case of a rising heart rate, the beam segment duration dt must be decreased toaccommodate the shorter quiescent interval. The dose dMU delivered during thattime must remain the same, however, so the dose rate dMUdt is increased. Therefore,the gantry rotation speed dθdt must be increased proportionally so that the dosedelivered throughout the arc segment dMUdθ remains the same. Since MLC leaf27displacement per degree of gantry rotation is described by the equationdxdθ=dxdt/dθdt(3.2)and must also remain unchanged, the leaf speed dxdt is also increased.3.1 Beam Delivery Time CalculationBeam delivery adaptation is not a trivial task, as a method for direct control of thebeam timing does not exist for the TrueBeam system. This limitation is stated inthe TrueBeam Developer Mode Manual:The XML beams do not contain a time element. In other words, itis not possible for the user to specify the exact time progression andoverall time required to deliver an XML beam.The linear accelerator will always deliver the plan as quickly as possible, the speedat which it progresses through the given control points dictated by whichever ma-chine parameter is the limiting factor.Recall that an XML beam is simply a specification of dose-position trajectory.The time progression is computed by the TrueBeam control system a priori (beforebeam on) when the beam is loaded and the prepare button is pressed. The controlsystem takes the XML position vs. MU trajectory and converts it into MU vs. timeand position vs. time trajectories, taking into account both the machine’s inherentlimitations (maximum dose rates, speeds, and accelerations) and any user-specifieddose rate ceilings or axis velocity limits.For each segment, the control system computes the dose rate and axis velocitiesbased on the number of MU to be delivered and the total axis movement. The lim-iting rate of the segment is first determined – either dose rate or maximum velocityof an axis. The rest of the segment parameters are then computed proportionallysuch that the position vs. MU trajectory is maintained. The limiting rate can alsovary from segment to segment. For example, the gantry rotation speed may bethe limiting factor for one segment, while the velocity of an MLC leaf may be thelimiting rate for the next.28The control system also contains algorithms to handle transitions between seg-ments (at control points). The inertia of the machine’s mechanical parts must betaken into account – in other words, instantaneous changes in axis velocities be-tween segments is not possible, so the control system must also consider accelera-tion of the mechanical axes and changes in dose rate.The total beam delivery time is simply the sum of the times required to delivereach segment. The control system automatically minimizes the total time by usingthe maximum possible rate for the limiting axis in each segment, providing no usercontrol over the beam delivery speed.3.2 Beam Timing Control Using Couch MovementDespite the lack of user control over beam timing in the TrueBeam system, amethod for indirect modification of the beam delivery speed was devised. Sincethe control system identifies the limiting rate for each segment and sets the otherparameters accordingly, manually determining the limiting axis and specifying therange of motion would, in theory, allow for manipulation of the time element.For example, if the longitudinal couch movement is forced to be the limitingaxis by setting the couch maximum velocity to 1 cm per second, moving the couch10 cm throughout one segment would cause that segment to take 10 seconds tocomplete.10cm1cm/s= 10 seconds (3.3)Unfortunately, velocity limits can only be set once (at the beginning of the XMLbeam) and cannot be changed mid-delivery. Therefore, the beam timing must becontrolled by varying the distances moved by the limiting axis.We chose to use the couch axis as the limiting rate since it is separate from thegantry and the beam-controlling components, and any modifications to the couchmovement would not alter the dose delivery.The couch method was first tested by changing the overall delivery time of a180 degree arc VMAT plan. Modified XML beams were created where the max-imum velocity for the longitudinal couch axis was set to 0.1 cm/s and the couchwas planned to move a certain distance from the first to the last control point.29Table 3.1: Control of overall beam time using couch movement.Couch Translation (cm) Expected Time (s) Actual Time (s)0 (original plan) As fast as possible 54.8610 10 / 0.1 = 100 100.2815 15 / 0.1 = 150 150.3020 20 / 0.1 = 200 200.24Table 3.1 shows the couch method timing experiment results obtained fromanalysis of the trajectory logs. The original plan, in which the couch remains sta-tionary, is delivered as quickly as possible – however, there is no user control overthe beam delivery time. Limiting the couch velocity to 0.1 cm/s and moving thecouch 10 cm over the course of the plan delivery results in an overall time of100.28 seconds. Similarly, moving the couch 15 and 20 cm results in overall de-livery times of 150.3 and 200.24 seconds, respectively. The similarity betweenthe expected and actual times demonstrates that the couch movement method canbe successfully used to control the time element, as long as the couch maximumvelocity is set low enough to be the limiting rate and the desired delivery time isgreater than the original delivery time. The actual times are slightly longer than theexpected times (within 0.3 seconds), most likely due to the additional time requiredto ramp up the dose rate at the beginning of the plan and to ramp down the doserate at the end of the plan.The following figures were created from the trajectory log results of the dif-ferent XML beams. Figure 3.1 illustrates the effectiveness of the couch method,increasing the time required to deliver the prescribed dose by moving the couch alarger distance. Greater couch translation essentially spreads out the beam deliveryover a longer time.30Figure 3.1: Cumulative MU over time for beam deliveries with 0, 10, 15, and20 cm couch translations.Similarly, Figure 3.2 shows the difference in control point progression overtime. The original plan is delivered as quickly as possible, while the couch-modified plans progress linearly through the control points at slower rates. If thecontrol point values are plotted against cumulative MU instead, we can see thateach beam still follows the planned position versus MU trajectory. In other words,the couch method works as intended – only the beam delivery time is affected,while the delivered treatment plan remains the same. Figures 3.4 and 3.5 showthe gantry angle versus time and MU, once again showing the differences in timingbut identical dose-position trajectories. Even with an axis that follows a more com-plicated path, such as MLC leaf 90, the position versus MU trajectory is properlymaintained (Figures 3.6 and 3.7).Our method of using couch translation to indirectly control the beam deliverytime was successful. The next step was to incorporate the technique into a planwith alternating beam-on and beam-off intervals, using couch movement to indi-vidually modify the beam interval durations with the goal of synchronizing thebeam delivery with a cardiac signal.31Figure 3.2: Control point progression over time for beam deliveries with 0,10, 15, and 20 cm couch translations.Figure 3.3: Control point vs. cumulative MU for beam deliveries with 0, 10,15, and 20 cm couch translations.32Figure 3.4: Gantry angle over time for beam deliveries with 0, 10, 15, and 20cm couch translations.Figure 3.5: Gantry angle vs. cumulative MU for beam deliveries with 0, 10,15, and 20 cm couch translations.33Figure 3.6: MLC leaf 90 position over time for beam deliveries with 0, 10,15, and 20 cm couch translations.Figure 3.7: MLC leaf 90 position vs. cumulative MU for beam deliverieswith 0, 10, 15, and 20 cm couch translations.34Chapter 4Creation of Interleaved PlansCardiac synchronized radiation treatment involves a beam delivery pattern of al-ternating beam-on and beam-off segments, precisely timed such that the beam-onsegments coincide with the quiescent intervals of the cardiac cycle. Adaptation ofa conventional VMAT plan to a cardiac signal requires consideration of the gantryrotation behaviour. The beam is on throughout the entire arc in a standard VMATplan delivery – beam-off segments must be introduced for the purpose of cardiacsynchronization. However, the plan delivery cannot be simply paused during thesesegments, as repeatedly stopping the gantry rotation would result in inaccurate dosedelivery due to the frequent and abrupt changes in momentum.254.1 Treatment Plan InterleavingWe used a treatment plan interleaving method to solve the problem of continu-ous gantry rotation and alternating beam-on and beam-off intervals. The originalVMAT plan is broken up into phases such that portions of the plan are assigned toeach phase in alternating fashion. As a result, individual phases are essentially sep-arate plans where the gantry rotates identically to the original plan, but the beamis turned on and off repeatedly. Each phase is responsible for different segmentsof the original treatment plan, and the sum of the dose deliveries from all phasesshould be equivalent to the original plan’s dose delivery.To create the interleaved plans, the original treatment plan must first be deliv-35ered on the TrueBeam machine so that the trajectory log may be collected. Thetrajectory log is then used to split the original plan into the desired number ofphases with the specified beam-on segment durations. Figures 4.1 and 4.2 show anexample of the process for creating two interleaved phases with 400 ms beam-onsegments. Using the control point versus time data from the original plan trajec-tory log, the first 400 ms segment is assigned to the first phase. The next 400ms segment is handled by the second phase, before the following segment is onceagain assigned to the first phase. The result is a beam delivery pattern of alter-nating 400 ms beam-on and beam-off intervals for each phase. The advantage ofthis method is that the gantry can continue to rotate during the beam-off intervals(vertical motion-only segments in Figure 4.2), as the original planned dose that is“missed” during these segments is delivered by the other phase. The individualphases exhibit the alternating beam-on and beam-off delivery pattern desired forcardiac synchronization, and when combined make up the original non-interleavedplan.Phase 1Phase 2Figure 4.1: The original plan is broken up into 400 ms portions, and eachportion is assigned to different phases in an alternating pattern to createthe interleaved plans.36Figure 4.2: The gantry angle vs. MU trajectory plot for an interleaved phase.Vertical segments are motion-only segments, where gantry rotation con-tinues but no MU is delivered.4.1.1 Interleaver CodeA Python module interleaver was created to handle the treatment plan in-terleaving. When the Interleaver object is created, the period (cardiac cycleperiod) and the number of phases are specified. The period is divided into shorterbeam-on segments according to the number of phases. The original plan DICOMfile must be specified so that the original DICOM-RT object can be copied andmodified. The original plan trajectory log must also be selected so that the beamdelivery timing information is provided. The main interleave method loopsthrough the trajectory log snapshots (instantaneous machine axis positions in 20ms intervals) and assigns control points to phases based on the provided timingspecifications. A Phase object is created to hold control point data for each re-spective phase, and XmlBeam objects (introduced in Section 2.4) are created totransfer this data to DICOM and XML beam files.For example, if two phases with 400 ms beam-on durations are to be created,the first Phase object is updated with axis data from the trajectory log snapshots37until a snapshot at a time greater than or equal to 400 ms is reached. The “current”phase number is incremented, and axis data is assigned to the second Phase objectuntil the snapshot at 800 ms is reached, at which point the phase number variable isreset. The data for each phase is then filtered to only keep control points where thebeam transitions between on and off states, as other control points merely containinterpolated axis data – this helps to reduce final file sizes and keeps the number ofcontrol points under the Developer Mode limit of 5000. Phases can then be writtento correctly-formatted DICOM and XML beam files, converting axis data to IECor Varian scale as necessary.4.2 Interleaving ResultsA 10X FFF VMAT plan (10 MV photons, flattening filter free) with a 180 degreearc and a dose rate of 2000 MU/min was interleaved using 2 phases with 500 msbeam-on segments. FFF mode was chosen so that high dose rates could be usedto minimize the overall treatment time as much as possible, since the addition ofbeam-off segments would increase the delivery time. Figure 4.3 shows the first10 seconds from the trajectory log of the original plan, where the dose rate is aconstant 2000 MU/min throughout the beam delivery.Note: The “actual” dose rate measurement is calculated by taking the differencebetween the cumulative MU values at every trajectory log sample point (20 msapart). In reality, the linear accelerator dose rate is not perfectly constant over suchshort time intervals, as illustrated by the fluctuations in Figure 4.3. Over a longerperiod of time, however, the average dose rate matches the planned dose rate. Theactual delivered dose for the entire plan is always within 0.1 MU of the expectedcumulative dose (a difference of less than 0.1% of the total delivered MU).Figure 4.4 shows the trajectory log results of the interleaved plan. While thealternating beam-on and beam-off delivery pattern was successful, the actual beam-on durations were much longer than expected. The time required for the dose rateto ramp up to the maximum and ramp down to 0 was about 300 ms each, which isunacceptable when the quiescent interval delivery window is typically about 400ms. In addition, the gantry is forced to come to a stop at every transition between380 2 4 6 8 10Time (s)0500100015002000Dose rate (MU/min)ActualExpectedFigure 4.3: Trajectory log results showing constant dose rate over time forthe original 2000 MU/min treatment plan (non-interleaved).beam-on and beam-off states (Figure 4.5). This limitation is mentioned in theDeveloper Mode manual, but a reason for this behaviour is not provided. Whenthe beam is turning on and off multiple times per second – as is the case withthe interleaved plans – the result is a very jerky gantry motion that is obviouslydetrimental to the dose delivery accuracy. This issue is also the cause of the longdose rate ramp times, as the control system performs a synchronized decelerationor acceleration of all moving axes and MU delivery at each of these transitions.Due to the velocity limits of the other mechanical axes such as the gantry andMLC leaves, the dose rate ramp speeds are automatically moderated in an attemptto maintain the planned dose delivery.39Figure 4.4: Trajectory log results for the first phase of a 2-phase interleavedplan with a dose rate of 2000 MU/min and 500 ms beam-on segments.Figure 4.5: Trajectory log results of the interleaved phase showing the gantryrotation behaviour in relation to the dose rate. The gantry is forced tostop at every instance where the beam turns on or off.40The effects of the dose rate ramp time issue were minimized by decreasing thetreatment plan dose rate to 800 MU/min. Figure 4.6 shows the dose rate over timefrom the non-interleaved 800 MU/min treatment plan. The plan was interleavedusing 2 phases and 400 ms beam intervals. Figure 4.2 is a plot of the dose rate overtime for the first interleaved phase, showing the successful beam-on and beam-off delivery pattern. The average full width at half maximum (FWHM) beam onduration was calculated to be 402 ms; however, the average full beam-on duration(from 0 MU/min to 0 MU/min) was still significantly longer (567 ms) than theplanned time. The large discrepancy between the planned and actual full beam-ondurations would undoubtedly cause complications for the cardiac synchronizationlater on.0 2 4 6 8 10Time (s)0200400600800100012001400Dose rate (MU/min)ActualExpectedFigure 4.6: Trajectory log results showing constant dose rate over time forthe original treatment plan (non-interleaved).410 2 4 6 8 10Time (s)0200400600800100012001400Dose rate (MU/min)ActualExpectedFigure 4.7: Trajectory log results for the interleaved phase planned for 400ms beam-on and beam-off segments.Although the lower dose rate helped with the beam-on segment duration issue,the jerky gantry motion was still unaddressed. Figure 4.8 shows both gantry po-sition and rotation speed over time for the interleaved plan. The control systemrecognizes the transitions in beam states and plans for the gantry deceleration, asseen by the dips to 0 deg/s in the figure. In reality, the considerable momentum ofthe heavy gantry results in a less than ideal motion, and the gantry actually rotatesin the opposite direction for brief moments.42Figure 4.8: Gantry angle and rotation speed over time for the 2-phase, 800MU/min, 400 ms beam-on interleaved plan. Instantaneous gantry stopswere still causing an undesirable jerky motion.The dose rate ramp and gantry motion issues were addressed with a cleverworkaround that was developed during testing of the cardiac synchronized beams –this will be discussed in Section 5.2. Prior to the breakthrough that solved the afore-mentioned issues, other solutions were used to improve the machine behaviour,such as increasing the time between beam-on segments to decrease the jerky mo-tion of the gantry.434.3 Arc Segment Timing ControlWith a method in place to create the alternating beam on and off interleaved plans,individual arc segments could be modified using the couch movement method de-scribed in Chapter 3. The XmlBeam class contains a method to set properties(machine axis positions) given initial and final values for a corresponding set ofcontrol points numbers. The values for the control points in between the givenones are linearly interpolated (otherwise the control system will interpret the XMLbeam data as a sudden jump in position). Along with an appropriate axis velocitylimit, this function is used to set the amount of couch longitudinal movement foreach arc segment, indirectly modifying the segment durations.Figure 4.9: Planned cumulative MU and couch longitudinal motion for abeam delivery with 2 second beam-off intervals. The couch moves 0.2cm at 0.1 cm/s to force each beam-off segment to last 2 seconds.An initial test of the individual arc segment timing control involved setting a2 second beam-off interval, while the beam-on segments were left unmodified at44Figure 4.10: Trajectory log results showing dose rate and couch position overtime for an interleaved phase with 2 second beam-off intervals (2phases, 800 MU/min, and 400 ms beam-on intervals). The averageactual beam-off interval duration was 1.99 seconds.400 ms. The couch longitudinal maximum velocity was set to 0.1 cm/s, and thecouch was planned to move 0.2 cm during every beam-off interval to achieve the 2second beam-off duration (0.2cm÷0.1cm/s = 2s).Figure 4.9 shows the planned couch motion in relation to the cumulative MU.Each motion-only segment (horizontal segments on the cumulative MU plot) isaccompanied by a 2 mm shift in longitudinal couch position. The couch remainsstationary during the beam-on segments. The movement direction is alternated sothat the couch remains in the same general area and does not drift towards thelimits.With the longer time between beam-on segments, the gantry rotation behaviourwas also more acceptable (Figure 4.11). Although still not ideal, the increased time45Figure 4.11: Trajectory log results showing gantry position and speed overtime for an interleaved phase with 2 second beam-off intervals (2phases, 800 MU/min, and 400 ms beam-on intervals).between the instantaneous gantry stops at each beam state transition resulted in asmoother motion overall. The frequency of gantry acceleration and decelerationwas decreased, with the gantry rotating more slowly during the beam-off segments.The ability to modify individual arc segment durations allowed for us to be-gin working on synchronizing beam delivery to a cardiac signal. Time control ofmotion-only segments was very accurate, but modification of the beam-on seg-ments was more complicated due to the dose rate ramps. Our initial attempts atcardiac synchronization will first be discussed, before the solution to gantry mo-tion and dose rate issues is described in Section 5.2.46Chapter 5Cardiac SynchronizationSynchronization of a linear accelerator beam delivery to a cardiac signal requiresconstant adjustment of the beam timing so that dose is delivered only during thequiescent intervals. Although the cardiac period is relatively consistent over a shortperiod of time, any change in heart rate would lead to a desynchronization of theplanned treatment and dose would be delivered outside of the quiescent intervals.Irradiating during the contraction phases of the cardiac cycle, when heart motion isnot at a minimum, would result in unwanted dose delivered to areas outside of thePTV.In practice, this beam timing adjustment would be done in real time, with anECG monitoring the patient’s heart rate. However, the current TrueBeam clinicalsystem does not allow for real-time plan adaptation. The entire plan’s position-dosetrajectory must be provided a priori, and all machine parameters and the beam de-livery timing are calculated by the control system before beam-on. Nevertheless,we demonstrated the feasibility of our cardiac synchronization technique by adapt-ing a conventional VMAT plan to a sample cardiac signal. The ECG informationis provided prior to beam delivery, but the beam timing adaptation and calculationsare carried out as if the cardiac signal was being transmitted in real time.475.1 Initial Attempt at SynchronizationA sample cardiac signal collected by an in-house developed ECG device was usedfor the cardiac synchronization tests.37 A Python module cardiacanalyzerwas created to handle the cardiac signal peak detection and beam synchronizationcalculations. The module contains a function to detect the R-peaks so that thecardiac cycle periods throughout the entire signal can be calculated (by taking thetime between R-peaks). Each R-peak is detected by looking for a change in signalamplitude that exceeds a threshold value, then identifying the maximum signalpoint within the surrounding segment as an R-peak. The original VMAT plan wasinterleaved using 2 phases with 300 ms beam-on intervals. The couch movementmethod could then be used to modify the beam-off times in order to synchronizethe beam delivery with the cardiac signal data. Initially, only the beam-off timeswere modified due to the beam-on interval dose rate ramp issues and the associatedtiming complications.5.1.1 Beam Timing CalculationThe cardiacanalyzer module includes a function to calculate the beam tim-ing in a way that simulates a real-time response to the cardiac period data. In thisinitial version, the beam was planned to turn on every other heart beat to avoid thejerky gantry motion associated with rapid changes in the beam state. Each beam-off duration was calculated based on the “predicted” cardiac period such that thenext beam-on segment would occur in the middle of a quiescent interval. The pre-dicted cardiac period was calculated by taking the average of the previous 2 cardiacperiods. In other words, the method simulates a real-time adaptation by using thelatest available cardiac period data to predict the next cardiac periods and plan thebeam timing accordingly.For example, the average of the first 2 cardiac periods of the sample signal istaken as the predicted cardiac period. The first beam-off duration (or start delay)is set so that the first beam-on occurs one third of a cardiac period after the lastR-peak. However, since the actual cardiac period may differ from the predictedperiod, the beam-on interval may, in reality, begin earlier or later than the idealtime. To ensure synchronization is maintained with a changing heart rate, the next48beam-off time is calculated relative to the last R-peak time and the end of the lastbeam on.Tbeam-off = tr-peak+(cardiac period×2)+ cardiac period3 − tbeam-on end (5.1)Equation 5.1 shows the calculation for the beam-off duration Tbeam-off, wheretr-peak is the time of the last R-peak and tbeam-on end is the time at which the lastbeam-on segment ended. The calculation used for the synchronized beam timing(beam-on every other heart beat) is also pictured in Figure 5.1.0 2 4 6 8 10Time (s)40020002004006008001000Doserate (MU/min)Dose rateCardiac signalLast R-peak Cardiac period * 2 Cardiac period / 3Beam-on endFigure 5.1: Average of the previous 2 cardiac periods (yellow) is taken asthe predicted cardiac period. The beam-off duration (blue) is calculatedrelative to the last R-peak and the predicted cardiac period, taking intoaccount the time of the last beam-on.A function in the interleaver module is then used to set the beam segmentdurations according to the timing calculated from the cardiac signal. The control49points of the interleaved phase are looped through, and beam-off segments areidentified by control point sections with 0 MU delivered. The couch translation foreach beam-off segment is set to achieve the planned segment duration determinedfrom the cardiacanalyzer code.An important detail to consider is that the couch longitudinal position is limitedto a 0.01 cm precision, according to analysis of the TrueBeam trajectory logs.Therefore, the couch maximum velocity must not be set too low or the timingprecision required for cardiac synchronization cannot be achieved. Initially, thecouch maximum velocity was set to 1 mm/s – this only allowed for a 0.1 secondtiming resolution (0.01cm÷0.1cm/s = 0.1s), which did not allow the TrueBeamsystem to properly follow the planned arc segment durations. However, the velocitylimit must still be set low enough to make the couch the limiting axis. Settingthe couch longitudinal maximum velocity to 1 cm/s instead (0.01 second timingresolution) was found to be adequate for the cardiac synchronized beam timing.5.1.2 Beam Timing CorrectionFollowing delivery of the cardiac synchronized beam, analysis of the trajectory logshowed that the beam delivery quickly becomes desynchronized (Figure 5.3). Thebeam-on segments drift outside of the quiescent intervals, delivering dose duringthe QRS portions of the cardiac signal. The actual beam cycle times (total timefor a beam-on and beam-off segment) failed to match the planned times due tothe beam-on segment durations being longer than anticipated. A correction of thebeam timing would be necessary to successfully deliver the cardiac synchronizedplan.To test a correction method, a simplified plan where the beam is turned on every2 seconds was created. An initial delivery of the plan and analysis of the trajec-tory log showed that the discrepancies between the actual beam segment durationsand the planned durations were not consistent throughout the plan. The beam-on segment timing, despite being untouched by any couch modifications, variedunpredictably. The actual beam cycle times from the initial delivery trajectory logwere compared with the planned times to determine the required corrections, whichwere applied to the beam-off times.50Figure 5.2: Comparison between beam cycle times for an initial plan deliv-ery and a corrected plan delivery. The beam cycle times (beam-on andbeam-off) were planned to be 2 seconds.Delivery of the corrected plan proved to be successful, as shown in Figure 5.2.An uncertainty of ±0.02 seconds from the trajectory logs is expected due to the 20ms sampling interval. Application of the correction method to the cardiac synchro-nized plan was also successful, and all beam-on segments remained between theR-peaks throughout the entire delivery (Figure 5.3).An artificial cardiac signal was created by replicating the data from one samplecardiac period and manipulating the data to gradually increase and then decreasethe heart rate. Our technique was tested with this cardiac signal to demonstrate theability to remain synchronized with a significantly changing heart rate (Figure 5.4).The beam timing correction solution, despite being able to achieve the desiredend result, was obviously not ideal. The entire process of delivering the originalplan, using the trajectory log to create the cardiac synchronized interleaved phases,then using the trajectory log of the initial deliveries to created corrected plans wasunnecessarily cumbersome. In an actual real-time application, an on-the-fly cor-rection to the beam timings may not be feasible. Fortunately, a better method was51Figure 5.3: Comparison of results between the uncorrected (top) and the cor-rected (bottom) cardiac synchronized plan. In the corrected plan de-livery, all beam-on segments are kept within the quiescent intervals asplanned.52Figure 5.4: Cardiac synchronization is maintained as the heart rate increases25 bpm. The dose rate plot shown is the corrected plan delivery.developed that solves the dose rate ramp and gantry motion issues, eliminating theneed for any timing correction. This new solution is described in the next section.5.2 Low Dose Beam-Off SolutionAs stated in the TrueBeam Developer Mode manual, every transition from a beamsegment (segment delivering MU) to a motion-only segment is accompanied byan instantaneous stop of all moving axes. In the case of our cardiac synchro-nized VMAT plans, the dose rate, gantry rotation speed, and MLC leaf speedsare decreased to zero in a synchronized manner. During a motion-only segment,machine axes are accelerated back to their maximum speeds (determined by thevelocity limits or the limiting axis). An instantaneous stop occurs once again at thetransition from a motion-only segment back to a beam segment.53Analysis of the trajectory logs revealed the presence of beam holds during ev-ery motion-only segment and at the beginning and end of the beam delivery. Evi-dently, the instantaneous stops are triggered at the beginning and end of every seg-ment the control system interprets as a beam hold. This makes sense for treatmentapplications where beam holds are traditionally used, such as in gating techniques.In respiration-gated radiotherapy, for example, the treatment is essentially haltedfor short segments of time during certain portions of the patient’s breathing cycle.The synchronized deceleration of all axes in proportion to the decrease in dose ratewhen the beam turns off is necessary so that accurate dose delivery is maintainedand treatment may resume as planned at the next beam-on. This technique is suit-able for lung gating due to the relatively longer respiratory period. Whereas theaverage respiratory period can range from 3 to 5 seconds,38 the cardiac period isoften less than one second. Rapid stopping and starting of the machine axes leadsto the jerky motion and inaccurate beam timing described in the previous section.To eliminate the beam holds throughout the delivery, the cardiac synchronizedplans were altered to deliver a negligible amount of dose (0.001 MU) during theprevious motion-only segments. This simple trick immediately solved all of thepreviously described issues. Only the beam holds at the beginning and end ofthe plan delivery remained – removal of the beam holds in between the beam-onsegments resulted in significantly reduced dose rate ramp times, as the dose rate nolonger needed to be synchronized with the accelerating or decelerating mechanicalaxes (Figure 5.5). For a dose rate ceiling of 800 MU/min, the old method withbeam holds required 140 ms to reach the maximum dose rate. The new low dosemethod allowed for the dose rate ramp time to be significantly reduced, reachingthe maximum dose rate in 40 ms (within 2 trajectory log snapshots).Perhaps most importantly, the low dose method allowed for the gantry to con-tinuously rotate instead of stopping and starting at the beginning and end of everybeam-on segment. Interleaving using 8 phases resulted in similar gantry rotationspeeds during both beam-on and beam-off segments, but some variation still ex-isted during the higher heart rate portion of the plan. Gantry rotation speed is de-pendent on the number of phases used for interleaving – a higher number of phaseswill result in larger gaps between beam-on segments for each phase, leading tofaster gantry rotation during the motion-only segments.54Figure 5.5: Comparison between dose rate ramp times calculated by theTrueBeam control system for the old method (top) and the new low dosemethod (bottom). For a maximum dose rate of 800 MU/min, the doserate ramp time was reduced from 140 to 40 ms (or within 2 trajectorylog snapshots).55Figure 5.6: Comparison between gantry rotation speed calculated by theTrueBeam control system for the old method (top) and the new low dosemethod (bottom). The original plan was interleaved using 8 phases.56As a result of the improvements to the dose rate ramp times and gantry smooth-ness, the beam delivery timing was also much more accurate. The actual times foreach beam-on segment matched the planned times without the need for any post-delivery correction (Figure 5.8).The new low dose method proved to be an effective solution to all of the pre-viously described issues caused by machine limitations. Delivering a negligible0.001 MU during each motion-only segment leads to an increase in overall treat-ment dose of less than 0.04%. In the next section, a refined technique of interleav-ing and cardiac synchronization is discussed. The new technique takes advantageof the low dose beam-off method and further improves the gantry rotation smooth-ness.Figure 5.7: No correction is needed for cardiac synchronization using the lowdose beam-off method. The reduced dose rate ramp times allow formore accurate beam timing.57Figure 5.8: Beam-on segment timing for the low dose method matches theplanned timing (indicated by the green lines) without any correction.5.3 Cardiac Synchronization with Smooth GantryMotionNew interleaving and beam timing techniques were used in combination with thelow dose method to reduce the inter-segment variations in gantry rotation speed. Toaccomplish this, it was important that both beam-on and beam-off segments weremodified proportionately so that the gantry rotation rate remained constant withineach cardiac period. The number of phases used for the interleaving process wasalso critical for achieving a smooth gantry motion.5.3.1 Dynamic Beam TimingWith the low dose method, the beam-on segment durations could be accuratelyset using the couch modification. During analysis of the cardiac signal and beam58timing calculation, each beam-on segment duration was planned to be one thirdof the predicted cardiac period. Since the gantry behaviour was also drasticallyimproved by the low dose method, the beam could be turned on every heart beatinstead of every other heart beat. Beam-off times were calculated similarly to thefirst method described in Section 5.1.1 (relative to the last R-peak and the lastbeam-on segment) but with a few differences. Only one previous cardiac periodwas used to predict the next period, the beam-on segments were assumed to lastone third of the cardiac period, and the beam was planned to turn on every beat.Tbeam-off = tr-peak+ cardiac period+cardiac period3− tbeam-on end (5.2)Last R-peak Cardiac periodCardiac period / 3Beam-on endFigure 5.9: The previous cardiac period (yellow) is used to calculate thebeam-on duration (red). The beam-off duration (blue) is calculated rela-tive to the last R-peak, taking into account the time of the last beam-on.595.3.2 Interleaving for Consistent Gantry Rotation SpeedChanging beam-on durations relative to the cardiac periods allows for proportionaladjustment of the beam-off durations. The optimal number of interleaved phasesdepends on the desired relative beam-on duration – interleaving using the correctnumber of phases will minimize changes in gantry rotation speed. For example, ifa beam-on segment is planned to last one third of the cardiac period and the beam isturned on every heart beat, the plan must be interleaved using 3 phases to maintaina consistent gantry rotation speed during the following beam-off segment. Figure5.10 helps explain the reasoning for the number of interleaved phases.Phase 2Phase 1 Phase 3Beam onBeam offPhase 2Phase 1Phase 3Beam onBeam offInterleavingPhase 1 DeliveryFigure 5.10: Interleaving using 3 phases allows for smooth gantry motionwithin a cardiac period when the beam is turned on every heart beat.Gantry speed is relatively consistent from beam-on to beam-off duringa phase delivery. The arc segment angular displacements are exagger-ated for illustration purposes.60For each predicted cardiac period, the beam is planned to be on for one third ofthe period. The following beam-off segment would then last roughly two thirds ofthe cardiac period (twice as long as the beam-on segment). Therefore, the gantry’sangular displacement during the beam-off segment must be twice the beam-on an-gular displacement so that the rate of rotation remains roughly the same.dθbeam-ondtbeam-on=dθbeam-offdtbeam-off(5.3)Using 3 phases for the plan interleaving achieves this – an arc segment from theoriginal plan is divided into 3 smaller arc segments, each with equal angular dis-placements and assigned to a different phase. During the delivery of one phase, thegantry rotates at a relatively consistent speed throughout each beam cycle (beam-onfollowed by beam-off), where the rotation speed depends on the predicted cardiacperiod.5.3.3 Cardiac SynchronizationThe couch movement beam timing method works by forcing the couch to be thelimiting axis – therefore, arc segment durations can only be increased from theoriginal period used for interleaving. It is important to use a reasonably shortperiod for the initial interleaving process so that the arc segment durations may beextended as necessary.We chose to use an interleaving period of 200 ms (the original plan was brokenup into 3 phases in 200 ms segments). Since the beam-on interval is set to onethird of a cardiac period, a 200 ms duration is suitable for a 600 ms cardiac period,which translates to a heart rate of 100 bpm. For lower heart rates, couch movementis used to lengthen the beam segment durations. As the heart rate decreases andthe cardiac period increases, the couch is translated greater distances to lengthenthe beam-on and beam-off segment durations. As the heart rate increases and thecardiac period decreases, couch movement is reduced to create shorter beam-onand beam-off durations, the shortest possible beam-on duration being the originalinterleaved period (200 ms).61Figure 5.11: Couch movement is increased or decreased in response to achanging heart rate, and dose rate is automatically adjusted. As theheart rate increases, couch translation is decreased to shorten the arcsegment durations, and the dose rate increases to maintain the planneddose delivery.Figure 5.11 shows the relation between couch translation distance and the arcsegment durations. Conveniently, the linear accelerator control system adjusts allother machine parameters according to the limiting axis, which in our case is al-ways the couch. The dose rate is automatically decreased when couch translationis increased in response to a lower heart rate, ensuring that the planned dose iscorrectly delivered over the extended period. In addition, the gantry rotation speedand MLC leaf speeds are also decreased accordingly. As the heart rate increases(cardiac period decreases), couch movement is decreased and the dose rate, gantryspeed, and MLC leaf speeds are automatically increased (Figure 5.11).Beam delivery of an interleaved phase was synchronized to the artificial car-diac signal with a rising and falling heart rate. The cardiac synchronization was62successful, as expected, and all beam-on intervals remained within the R-peaks asthe heart rate varied.Figure 5.12: Beam delivery for a 3-phase interleaved plan successfully syn-chronized to an increasing heart rate. All beam-on intervals are keptwithin the R-peaks.The new dynamic beam timing and 3-phase interleaving methods resulted inthe smoothest possible gantry motion. There was no significant acceleration ordeceleration of the gantry between beam-on and beam-off segments at any pointthroughout the beam delivery. Gantry rotation speed was only increased and de-creased in response to the changing heart rate, as desired.63Figure 5.13: Comparison between gantry rotation speed calculated by theTrueBeam control system for a 3-phase interleaved plan using the oldmethod (top) and the new low dose method with dynamic beam-ontimes (bottom). The gantry rotation speed changes in response to thechanging heart rate, but the gantry motion remains smooth between arcsegments.64Chapter 6Treatment Accuracy VerificationVerification of the dose delivery accuracy for the cardiac synchronized beam wasdone using film dosimetry. Since the treatment couch constantly moves back andforth during delivery of the cardiac synchronized beam, film could not be placeddirectly on top of the couch. A custom table was built that allowed for the couch tomove freely underneath while the phantom or film could be placed on the stationarytable surface. The table was made height-adjustable so that the phantom or filmcould be positioned as desired and to ensure the treatment couch could safely fitbetween the brackets (Figure 6.1). Two separate films were irradiated using theoriginal treatment plan and the cardiac synchronized phases, then compared usingthe FilmQA Pro software (Ashland Advanced Materials).6.1 Measurement Equipment and Treatment PlanningThe body-shaped oval portion of a QUASAR respiratory motion phantom (ModusMedical) was used for the treatment accuracy measurements (shown in Figure 6.4).This water-equivalent phantom contains 2 cavities in which cylindrical inserts canbe placed. A film cassette, which holds the film and punctures it at 3 corners tocreate fiducials, was inserted in the centre cavity. A cedar insert, which representslung density, was inserted in the outer cavity.Partial arc VMAT (0 to 180 degrees) is used for lung cancer radiotherapy tominimize unnecessary dose deposition to normal structures.39 For the cardiac syn-65Figure 6.1: A custom table was built for treatment accuracy verification of thecardiac synchronized beam. The table allows for the treatment couch tomove freely underneath in the longitudinal direction, while a phantomcan be placed on the table surface.chronized radiotherapy tests, however, the treatment had to be planned with thegantry rotating from 270 to 90 degrees (clockwise) to avoid collision with the cus-tom table. A polystyrene foam holder was constructed to hold the phantom in anupright position, simulating a patient lying on his side. A CT scan of the phantomin this position was acquired for creation of an SRS VMAT plan in the EclipseTPS. A dose of 4 Gy was prescribed to the PTV, requiring a machine output of1296.1 MU for a 10X FFF beam energy and 2400 MU/min dose rate.66Figure 6.2: An SRS VMAT plan for the film measurement was created in theEclipse TPS. The PTV (red outline) and the planned gantry rotation areshown.Figure 6.3: The isodoses for the treatment plan from the top-down view.67Figure 6.4: A QUASAR phantom with a wood insert and film cassette wasused for the treatment accuracy verification. A polystyrene foam blockwas used to hold the phantom in the planned treatment position and aCT scan was taken for treatment planning.6.2 Film MeasurementThe original plan was first delivered without the table or phantom in place, andthe trajectory log was used to create 3 interleaved phase XML beams. The phaseswere synchronized to the artificial cardiac signal with a rising and falling heart rate(ranging from 49 to 78 bpm).6.2.1 Treatment Plan DosimetryA 6.8 cm by 11 cm strip of Gafchromic EBT3 film was inserted in the QUASARphantom film cassette. The phantom was placed in the polystyrene foam block andon top of the custom table. The film was positioned at the isocentre by adjusting68the table height and aligning the crosshairs on the phantom sides with the roomlasers. A level was used to adjust the film cassette rotation and ensure the film wasparallel with the table. The original treatment plan was then delivered, overridingall couch parameters other than the longitudinal position so that the couch setupwithin the custom table would not be disturbed.Following delivery of the original plan and removal of the irradiated film, asecond 6.8 cm by 11 cm film strip was placed in the phantom insert. The secondfilm strip was irradiated by all 3 cardiac synchronized phases, which were deliveredone after the other in developer mode.Figure 6.5: The phantom and custom table were used to take film measure-ments for both the original treatment plan and the cardiac synchronizedinterleaved plan.696.2.2 Calibration CurveCalibration films were created by delivering 7 different known doses to films in asolid water setup. Each film strip was placed at 100 cm SAD with 5 cm depth and10 cm of backscatter (Figure 6.6). A 10X FFF beam with a 2400 MU/min dose rateand a 10 cm by 10 cm field size was used to irradiate each film. The TrueBeamlinear accelerator used for the measurements was calibrated to deliver 1 cGy perMU at a depth of 5 cm for a 10X FFF beam, making it simple to deliver the desireddoses to the films.The asymptotic fitting function used for calibration in the FilmQA Pro soft-ware works best when the calibration doses increase in geometric progression.40The highest calibration dose must also be greater than the highest expected dosedelivered to the treatment films. The maximum dose reported in the Eclipse TPSwas 138.3 percent of the prescribed dose, or 553.2 cGy. Thus, doses of 650, 433.3,288.9, 192.6, 128.4, 85.6, and 57.1 cGy were delivered to the calibration films.One extra film was left unexposed.Figure 6.6: Solid water setup for the calibration curve film measurements.The films were placed at 100 cm SAD with 5 cm depth and 10 cmbackscatter. A 10X FFF beam energy and a 10 cm x 10 cm field sizewere used.706.2.3 Film AnalysisAll films were scanned at least 24 hours after irradiation to minimize uncertaintiesassociated with post-exposure changes in optical density.41 The films were placedin the same orientation on an Epson Expression 10000XL scanner and scannedusing the recommended settings for film: 48-bit colour, 150 dpi resolution, and allcolour correction disabled. A 650 cGy reference film and an unexposed film wereincluded in the treatment film scans.Using FilmQA Pro, a calibration curve was constructed from the known doseregions of the calibration films. The reciprocal function X(D) = a+b/(D−c) wasused for the calibration curve fitting, where X(D,n) is the scanner response in thenth colour channel for an absorbed dose D and a, b, and c are constants.Figure 6.7: Films used for the calibration curve. From left to right: 650,433.3, 288.9, 192.6, 128.4, 85.6, 57.1, and 0 (unexposed) cGy.Dose maps were calculated for the treatment film scans, using the referencefilms to re-scale the calibration function for each scan. After marking the fiduciallocations for both treatment films, the scans were overlaid (matching the fiduciallocations) to compare the dose distributions between the original and interleavedtreatments.71Figure 6.8: Film scan for the original treatment plan delivery. A 650 cGy filmand an unexposed film were used as reference films.Figure 6.9: Film scan for the cardiac synchronized interleaved plan delivery.A 650 cGy film and an unexposed film were used as reference films.72The red channel was used for the dose analysis, as it has the greatest dose re-sponse up to 10 Gy.42 Gamma analysis, which combines distance-to-agreement(DTA) and dose difference (DD) factors, was used to compare the dose distribu-tions. DTA is the closest distance from a point on the first dose distribution to apoint on the second distribution with the same amount of dose. DD is the percentdose difference between a point on the first dose distribution and the same pointon the second distribution. Since both the dose difference and misalignment con-tribute to the difference of 2 distributions, the gamma index γ (Equation 6.1) isused for a complete dose distribution analysis.43γ(r1) = min{Γ(r1,r2)}∀{r2} (6.1)whereΓ(r1,r2) =√(r1− r2)2∆d2+[D1(r1)−D2(r2)]2∆D2(6.2)The gamma index for each point on a dose distribution is calculated by minimizingthe Γ function in Equation 6.2, where r1 and r2 are the vector positions of pointson the first and second dose distributions, D1(r1) and D2(r2) are the doses at thosepoints, and ∆d2 and ∆D2 are the DTA and DD criteria, respectively. The gammapassing criterion is γ(r1)≤ 1.A gamma passing rate of 99.4% was given (2%/2mm tolerance) for the dosedistribution comparison between the original and interleaved treatments. The iso-dose comparison is shown in Figure 6.10. A line profile was also drawn horizon-tally across the centre of the irradiated area on the overlaid films to compare thedose profiles (Figure 6.11). Overall, analysis of the films showed a remarkableagreement between the 2 dose distributions, demonstrating the effectiveness of thecardiac synchronized beam delivery technique.73Figure 6.10: FilmQA Pro dose isolines comparing the original treatment planand the cardiac synchronized interleaved plan.74Figure 6.11: FilmQA Pro dose profile comparing the original treatment plan(light red) and the cardiac synchronized interleaved plan (dark red).The line profile was drawn horizontally across the centre of the irradi-ated area.75Chapter 7ConclusionSuccessful synchronization of the interleaved phases and verification of the overalldose distribution accuracy demonstrates the feasibility of a cardiac synchronizedVMAT technique. Although hardware limitations on the TrueBeam linear acceler-ators prevented real-time plan adaption, the technique was tested by synchronizingthe beam delivery to ECG data available a priori.Table 7.1 summarizes the beam delivery times for the original treatment plan(non-synchronized) and the interleaved phases (cardiac synchronized).Table 7.1: Beam delivery time comparison.Treatment Plan Delivery Time (s)Original plan 97.6Phase 1 154.42Phase 2 153.66Phase 3 153.62Interleaved phase total 461.7The increased total delivery time for the cardiac synchronized phases (7.7 min-utes) is a compromise for the potential to minimize dose to surrounding tissuesusing the cardiac synchronization technique. If the original plan were deliveredusing a traditional gating technique, the beam would need to be turned on and off76488 times to deliver the entire plan in 200 ms beam-on intervals. Without the in-terleaving method, the dose rate changes would be accompanied by rapid gantryacceleration and deceleration, which would not only increase the delivery time butalso decrease the dose delivery accuracy. In contrast, the interleaved cardiac syn-chronization method allows for smooth gantry motion and utilization of a cardiacperiod prediction algorithm.The film dosimetry measurements presented in the previous chapter show thatthe combined interleaved phases – each synchronized to a cardiac signal – accu-rately deliver the prescribed dose. Despite significant modification to the originaltreatment plan to achieve the desired beam delivery behaviour, the measured dosedistributions for the original plan and the cardiac synchronized phases were re-markably similar. Overall, the results from this thesis work are a promising firststep towards the development of a cardiac synchronized VMAT technique that canbe an effective treatment for cardiac arrhythmias.7.1 Future WorkComparison with Traditional Radiosurgery TechniquesFurther experiments should be carried out to compare dose differences betweenthe cardiac synchronized technique and non-synchronized techniques. A cardiacradiosurgery treatment where beam delivery is not adapted to the cardiac signalwould require a larger PTV to account for heart motion. Non-synchronized andsynchronized treatment plans should be created to deliver a prescribed dose to thesame moving target, and the dose delivered to the target and to the surroundingareas should be evaluated.Real-Time Calculation from Live ECG DataThe beam timing adjustments for the cardiac synchronized plans discussed in thisthesis, although calculated in a manner that simulates real-time adaptation, aredetermined prior to the actual beam delivery using a complete cardiac signal. In-trafraction adjustment is dependent on the linear accelerator hardware, but beamtiming calculation in real-time using live ECG data is certainly achievable. The in-77house developed heart rate monitoring device may be interfaced with to producetiming parameter adjustments in true real-time, and successful cardiac synchro-nized beam delivery should be demonstrated. The device also monitors respiratorymotion simultaneously – this data should eventually be incorporated into the syn-chronization technique.Optimization TechniqueThe current technique of adapting a conventional VMAT plan for cardiac syn-chronization requires splitting the treatment into multiple phases. Developmentof a new optimization technique that explicitly incorporates heart motion would beideal. Linear accelerator parameters such as dose rate, MLC leaf positions, andgantry rotation will be optimized taking into account the periodic nature of the car-diac cycle, with the goal of increasing efficiency and decreasing overall treatmenttime.78Bibliography1. Ewer MS, Yeh E. Cancer and the Heart. Hamilton, ON: BC Decker Inc;2006. → page 12. Harris P, Lysitsas D. Ventricular arrhythmias and sudden cardiac death. BJAEduc. 2016;16:221–229. → page 13. Watchie J. Cardiovascular and Pulmonary Physical Therapy: A ClinicalManual. St. Louis, MO: Saunders Elsevier; 2009. → page 14. Gatzoulis MA, Balaji S, Webber SA, et al. Risk factors for arrhythmia andsudden cardiac death late after repair of tetralogy of Fallot: a multicentrestudy. Lancet. 2000;356:975–981. → page 15. Von Olshausen K, Witt T, Schmidt G, Meyer J. Ventricular tachycardia as acause of sudden death in patients with aortic valve disease. Am J Cardiol.1987;59:1214–1215. → page 16. Baye´s De Luna A. Clinical Electrocardiography. Oxford, UK:Wiley-Blackwell; 2012. → page 17. Bhatt N, Turakhia M, Fogarty TJ. Cost-effectiveness of cardiac radiosurgeryfor atrial fibrillation: implications for reducing health care morbidity,utilization, and costs. Cureus. 2016;8:e720. → pages 1, 68. Colilla S, Crow A, Petkun W, Singer DE, Simon T, Liu X. Estimates ofcurrent and future incidence and prevalence of atrial fibrillation in the U.S.adult population. Am J Cardiol. 2013;112:1142–1147. → page 19. Savelieva I, Camm J. Update on atrial fibrillation: part I. Clin Cardiol. 2008;31:55–62. → page 110. Mehra R. Global public health problem of sudden cardiac death. JElectrocardiol. 2007;40:S118–S122. → page 17911. Thiriet M. Anatomy and Physiology of the Circulatory and VentilatorySystems. New York, NY: Springer New York; 2014. → page 212. Kennedy A, Finlay DD, Guldenring D, Bond R, Moran K, McLaughlin J.The cardiac conduction system: generation and conduction of the cardiacimpulse. Crit Care Nurs Clin North Am. 2016;28:269–279. → page 213. Issa ZF, Miller JM, Zipes DP. Clinical Arrhythmology andElectrophysiology: A Companion to Braunwald’s Heart Disease. 2nd ed.Philadelphia, PA: Saunders Elsevier; 2012. → page 214. Gaztan˜aga L, Marchlinski FE, Betensky BP, Gaztan˜aga L, Marchlinski FE,Betensky BP. Mechanisms of cardiac arrhythmias. Rev Esp Cardiol. 2012;65:174–185. → page 315. Mines GR. On circulating excitations in heart muscle and their possiblerelation to tachycardia and fibrillation. T Roy Soc Can. 1914;8. → page 316. Zei PC, Soltys S. Ablative radiotherapy as a noninvasive alternative tocatheter ablation for cardiac arrhythmias. Curr Cardiol Rep. 2017;19. →page 317. Ernst S. Catheter ablation. Card Electrophysiol Clin. 2017;9:311–317. →pages 4, 518. de Jong S, van Veen TAB, van Rijen HVM, de Bakker JMT. Fibrosis andcardiac arrhythmias. J Cardiovasc Pharm. 2011;57:630–638. → page 419. Baye´s de Luna A. Basic Electrocardiography: Normal and Abnormal ECGPatterns. Malden, MA: Blackwell Pub; 2007. → page 420. Fogoros RN. Antiarrhythmic Drugs : A Practical Guide. Malden, MA:Blackwell Pub; 2007. → page 521. Wyse D, Waldo A, DiMarco J, et al. A comparison of rate control and rhythmcontrol in patients with atrial fibrillation. N Engl J Med. 2002;347:1825–1833. → page 522. Ghzally Y, Gerasimon G. Catheter Ablation. In: StatPearls. Treasure Island,FL: StatPearls Publishing; 2018. → page 523. Gianni C, Mohanty S, Trivedi C, et al. Alternative approaches for ablation ofresistant ventricular tachycardia. Card Electrophysiol Clin. 2017;9:93–98. →page 58024. Shah RU, Freeman JV, Shilane D, Wang PJ, Go AS, Hlatky MA. Proceduralcomplications, rehospitalizations, and repeat procedures after catheterablation for atrial fibrillation. J Am Coll Cardiol. 2012;59:143–149. → page 525. Otto K. Volumetric modulated arc therapy: IMRT in a single gantry arc:single arc radiation therapy. Med Phys. 2007;35:310–317. → pages 6, 8, 3526. Chin LS. Principles and Practice of Stereotactic Radiosurgery. New York,NY: Springer; 2008. → page 627. Simeonova AO, Fleckenstein K, Wertz H, et al. Are three doses of stereotacticablative radiotherapy (SABR) more effective than 30 doses of conventionalradiotherapy? Transl Lung Cancer Res. 2012;1:45–53. → page 628. Soltys SG, Kalani MYS, Cheshier SH, Szabo KA, Lo A, Chang SD.Stereotactic radiosurgery for a cardiac sarcoma: a case report. TechnolCancer Res Treat. 2008;7:363–367. → page 629. Bhatt N, Fogarty T, Maguire P. Cardiac radiosurgery for the treatment ofatrial fibrillation. World J Cardiovasc Dis. 2016;06:143. → pages 6, 730. Cuculich PS, Schill MR, Kashani R, et al. Noninvasive cardiac radiation forablation of ventricular tachycardia. N Engl J Med. 2017;377:2325–2336. →pages 6, 731. Ramanathan C, Jia P, Ghanem R, Calvetti D, Rudy Y. Noninvasiveelectrocardiographic imaging (ECGI). Ann Biomed Eng. 2003;31:981–994.→ page 632. Barnes EA, Murray BR, Robinson DM, Underwood LJ, Hanson J, Roa WHY.Dosimetric evaluation of lung tumor immobilization using breath hold at deepinspiration. Int J Radiat Oncol Biol Phys. 2001;50:1091–1098. → page 733. Wang X, Pan T, Pinnix C, et al. Cardiac motion during deep-inspirationbreath-hold: implications for breast cancer radiotherapy. Int J Radiat OncolBiol Phys. 2012;82:708–714. → page 734. Gavaghan M. Cardiac anatomy and physiology: a review. AORN Journal.1998;67:800–822. → page 735. Sharma SD. Unflattened photon beams from the standard flattening filter freeaccelerators for radiotherapy: advantages, limitations and challenges. J MedPhys. 2011;36:123–125. → page 128136. Zaila A, Adili M, Bamajboor S. Pylinac: a toolkit for performing TG-142 QArelated tasks on linear accelerator. Phys Medica. 2016;32:292–293. → page2037. Kohli K, Liu J, Schellenberg D, et al. Prototype development of an electricalimpedance based simultaneous respiratory and cardiac monitoring system forgated radiotherapy. Biomed Eng OnLine. 2014;13:144. → page 4838. Seppenwoolde Y, Shirato H, Kitamura K, et al. Precise and real-timemeasurement of 3D tumor motion in lung due to breathing and heartbeat,measured during radiotherapy. Int J Radiat Oncol. 2002;53:822–834. →page 5439. McGrath SD, Matuszak MM, Yan D, Kestin LL, Martinez AA, Grills IS.Volumetric modulated arc therapy for delivery of hypofractionatedstereotactic lung radiotherapy: a dosimetric and treatment efficiency analysis.Radiother Oncol. 2010;95:153–157. → page 6540. Lewis D, Micke A, Yu X, Chan MF. An efficient protocol for radiochromicfilm dosimetry combining calibration and measurement in a single scan. MedPhys. 2012;39:6339–6350. → page 7041. Khachonkham S, Dreindl R, Heilemann G, et al. Characteristic of EBT-XDand EBT3 radiochromic film dosimetry for photon and proton beams. PhysMed Biol. 2018;63:065007. → page 7142. Borca VC, Pasquino M, Russo G, et al. Dosimetric characterization and useof GAFCHROMIC EBT3 film for IMRT dose verification. J Appl Clin MedPhys. 2013;14:158–171. → page 7343. Low DA, Harms WB, Mutic S, Purdy JA. A technique for the quantitativeevaluation of dose distributions. Med Phys. 1998;25:656–661. → page 7382


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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