Management and Modeling for the Reduction of
the PTV Margin in EBRT
Rangika Bernadette Abeygunasekera
S u b m itte d for th e D egree of M aster of P h ilosophy
from th e U niversity of S urrey
C entre for V ision, Speech an d Signal P rocessing F aculty of E ngineering an d P h ysical Sciences
U niversity of S urrey G uildford, S u rrey GU2 7XH, U.K.
JuFy 2013
ProQ uest N um ber: 27558187
All rights r e se r v e d
INFORMATION TO ALL USERS
The q u a lity of this r e p r o d u c tio n is d e p e n d e n t u p o n the q u a lity of the c o p y s u b m itte d .
In the unlikely e v e n t that the a u th o r did not sen d a c o m p l e t e m a n u scrip t
and there are m issing p a g e s , th e se will be n o t e d . Also, if m a teria l had to be r e m o v e d , a n o t e will in d ic a te the d e le t io n .
uest
P ro Q u est 27558187
Published by ProQ uest LLO (2019). C o p y rig h t of the Dissertation is held by the Author.
All rights r eserv ed .
This work is p r o te c te d a g a in s t u n a u th o riz ed c o p y in g under Title 17, United S ta tes C o d e M icroform Edition © ProQ uest LLO.
ProQ uest LLO.
7 8 9 East Eisenhow er Parkway P.Q. Box 1346
E xternal Beam R adiotherapy is a m ethod th a t is widely used in treating tum ours in cancer patients. Image acquisition m ethods such as 4 - dimensional com puted tom ogra phy (4DCT) and cone-beam computed tom ography(CBCT) are used frequently when planning and delivering treatm ent strategies. However, patients involuntary movements such as respiratory motion cause ambiguities (blurriness, etc ) when locating tum ours in these scanned images. These aforementioned ambiguities in tum our location could lead to higher dosed treatm ent plans. This could be highly undesirable for the patients, as it will easily damage healthy tissues around th e tum our as well. However, two m ain problems occur in external beam radiotherapy and need to be taken into account when estim ating th e motion of a tum our. First, th e tum our movement as the patient breathes and second, the delay in the tim e needed for th e treatm en t delivery system (usually a linear accelerator) to move and produce a beam profile th a t conforms to th e treatm en t plan at a particular angle. This means th a t ideally the patients internal organ configu ration should be dynamically predicted in order to minimize treating norm al tissue a t the tum our margin.
We propose th a t a dynamic CT volume (one breathing cycle) be used, to determ ine a patient specific motion model by using iterative registration m ethods to determ ine the transform ation param eters to locate th e tum our and its movements. A correlation model was built to determine the motion between the lung tum our and the internal surrogate organ. Therefore, when the treatm ent beams impinge on the patient during radiotherapy the tum our position can then be determ ined and targeted w ithin ± 3m m from the C TV -PTV margin in any direction which accounts for geometric errors in treatm ent planning. As a consequence, reduces the overall P T V margin of error in tum our motion considering imperceptible penetration, which leads to reduced patien t dose in the treatm ent plan.
K e y w o rd s: Radiotherapy, Respiratory-induced Motion, Image Registration, Lung tum our margins. Modeling, Planning Target Volume (PTV ) margin.
Email : r . abeygunaseker a@ee .surrey.ac.uk WWW : http://w w w .eps.surrey.ac.uk/
A cknow ledgem ents
I would like to thank my parrot Archibald for all of his kind advice. I would like to th an k my supervisors. Very specially to Dr. Em m a Lewis for her kindness, under standing, advice and support over the years. At times you were more like a friend than a supervisor. I appreciate very much for giving me a topic th a t interested me from the beginning of my research. I would also like to than k Dr. Kevin Wells for giving me the opportunity to carry out my research. I would like to th ank everyone in CVSSP and all my friends Pve made in and out of CVSSP over the many years in Surrey.
Finally, I would like to give a very big thank you to everyone in my family for their continuous love and support over the years. Even when times were tough you were right behind me supporting me and I thank you all for it. I would like to give a very special than k you to my mum, especially for her unconditional love and support all throughout my life. Mum, if it weren’t for your faith I wouldn’t even be here today. You always believed in me and I thank you for it. Dad, I would like to th an k you too for your love and support and for giving me th e opportunity to study in UK. Last b u t not least I would like to thank my one and only sister, for her love, support and friendship. I would not have got this far in life if all of you weren’t p art of my life and so I am eternally grateful.
1 In tr o d u c tio n 1
1.1 Thoracic and Abdominal C a n c e r ... 2
1.2 Diagnostic Im a g in g ... 3
1.3 R a d io th e r a p y ... 4
1.3.1 Internal R a d io th e ra p y ... 4
1.3.2 Systematic Radioisotope T h e ra p y ... 6
1.3.3 External Beam R a d io th e r a p y ... 6
1.3.4 Treatm ent Planning and Treatm ent D e l i v e r y ... 7
Delineation of Tumour M a rg in s ... 8
1.3.5 L im ita tio n s ... 9
1.4 State-of-the-A rt Imaging and Guidance ... 11
1.4.1 V irtual s im u l a ti o n ... 12
3D-Conformal R ad io th erap y ... 12
Intensity M odulated R a d io th e r a p y ... 12
Image Guided R adiotherapy ( I G R T ) ... 12
1.5 Motivation ... 13
1.6 Proposed M e th o d ... 14
1.7 Outline of T h e s i s ... 15
1.8 A ch iev e m en ts... 15
2 R esp ira to ry -in d u ced T um our M o tio n an d M o tio n M o d e llin g in R a d io th era p y 17 2.1 Physiology of the Respiratory System ... 18
2.2 Mechanics of Respiration ... 18
vi Contents 2.2.1 I n h a la tio n ... 19 2.2.2 Exhalation ... 20 2.2.3 H y ste re s is ... 21 2.3 Respiratory Motion in R a d io th e ra p y ... 22 2.3.1 Measuring Respiratory M o t i o n ... 25 2.4 Respiratory-induced Motion M a n a g e m e n t... 30 2.4.1 Treatm ent P l a n n i n g ... 30 2.4.2 Quality A s s u r a n c e ... 31
2.5 EBRT M ethods Accounting for Respiratory-induced M o tio n ... 31
2.5.1 Motion Encompassing M e th o d s ... 32
Slow CT S c a n n in g ... 32
Inhale-Exhale Breath-hold CT ... 33
2.5.2 Modelling Respiratory-induced Motion to A dapt P atient Specific Motion ... 34
2.6 S u m m a r y ... 34
3 T rea tm en t P lan n in g: R esp ira to ry -in d u ced M o tio n E x tra ctio n 37 3.1 Tum our Localisation M e th o d s ... 38
3.2 Image D ata for Respiratory M o tio n ... 40
3.2.1 4-Dimensional Computed Tomography (4DCT) Image D ata . . . 40
3.2.2 Simulated D ata: 4D X C A T ... 41
3.3 Segmentation of Organs Affected by Respiratory-induced Motion . . . . 43
3.4 Image R e g is tra tio n ... 45
3.4.1 Spatial Transformation M o d e ls ... 51
3.4.2 Rigid Transformation Model ... 53
3.4.3 Affine Transformation M o d e l ... 54
3.5 Iterative Closest Point M e t h o d ... 55
3.6 E xtraction of Respiratory Motion from Simulated D ata ... 57
3.7 Validation of the Registration A lg o r ith m s ... 60
4 R esp ira to ry -in d u ced M o tio n C o rrelation M o d e l 63
4.1 Respiratory-induced Motion Reduction in D im e n s io n a lity ... 65 4.2 Principal Component Analysis ... 65 4.3 Canonical Correlation A n a ly s is ... 68 4.4 Evaluation of the Correlation of Respiratory-induced M otion Models . . 71
4.4.1 Analysis on the Lung Tumour M o t i o n ... 71 4.4.2 Analysis of the Motion Between External Surface M arkers and
Lung T u m o u r s ... 72 4.4.3 Analysis of the Internal Surrogate Organs and the Lung Tum our 74
Analysis of Respiratory-induced Motion Between Lung and Lung T u m o u r... 76 Analysis of Respiratory-induced Motion Between Liver and Lung
T u m o u r ... 82 4.5 S u m m a r y ... 86
5 Im p lem e n ta tio n o f th e R esp ira to ry -in d u ced M o tio n C o rrela tio n M o d e l 89
5.1 Analysis of the Extracted R espiratory M o tio n ... 91 5.2 Evaluation of the Lung Tumour Motion and Breathing P a t t e r n ... 96 5.3 Motion Correlation Model between the External Surface M arkers and
the Lung T u m o u r... 99 5.4 Motion Correlation Model between Lung Tum our and Internal Surrogate 101
5.4.1 Evaluation of Motion between the Lungs and Lung Tum our . . . 102 5.4.2 Evaluation of Motion between the Liver and Lung Tum our . . . 105 5.4.3 Evaluation of Respiratory-induced Correlation Model Between
Internal Surrogate and Lung Tumour M o t i o n ...108 5.5 Feasibility w ith Different P a tie n ts ... 109 5.6 S u m m a r y ... 110
6 C o n clu sio n s and F u tu re W ork 113
6.1 Concluding D is c u s sio n ...113 6.2 Future Work ... 115
viii Contents B R a d io th e r a p y T r e a tm e n t M a c h in e s 119 B .l Linear A c c e le r a to r ...119 B.2 M ulti-Leaf Collimator ( M L C ) ... 121 C T a b u la te d R a w D a ta 123 B ib lio g r a p h y 125
1.1 Cancer statistics for th e year 2010 as issued by the cancer research UK
[122] 2
1.2 Illustration of the operation of a CT scanner ... 4 1.3 Interstitial Brachytherapy procedure performing the seed im plantation
on a prostate cancer patient, using a needle [83]... 5 1.4 Illustration of the external beam radiation therapy performed on a pa
tient w ith pleural mesothelioma cancer (cancer attacking th e membranes of the heart, lungs and the abdom inal cavity) [129]... 7 1.5 Schematic illustrating the process of treatm ent planning: (a) CT image
illustrating the tum our on th e right lung and is used as reference to delin eate th e target volume margins, (b) Zoomed in on the area where target volume margins are delineated to show th e G TV (red line), extending for microscopic spread CTV (green line) margin, and finally P T V (yel low line) accounting for intra-fraction errors and to which th e dose is prescribed. Treated volume margin (purple line) and finally the irradi ated volume margin in (blue line) according to the ICRU report 50 [84]. [Note: CT image th a t was used to delineate was obtained from patient 1 of P O P I [123] d ata set]... 9 1.6 Illustration of the full motion p ath of the tum our and its margins due
to respiration... 10 1.7 Schematic representation of th e proposed m ethod. A dynam ic CT vol
ume will be used to param eterise the tum our location and subsequently tum our motion is observed in order to determ ine the tum our motion pathway w ith respect to the p atien t’s breathing pattern. This is then used to determine the entire tum our motion p ath and subsequently re duce the P T V margin for dose delivery. ... 16
2.1 Illustration of the anterior view of the structure of the respiratory system
[5 3 ]... 19
2.2 Physiology of typical healthy lungs during inhalation [29]... 20 2.3 Trajectories of the hysteresis c u r v e ... 22
List o f Figures
2.4 Stages of motion c o m p e n s a tio n ... 24 2.5 Illustration of the coronal views of the aforementioned blurriness of the
acquired diagnostic images from the same patient due to respiratory in duced motion in the thoracic and abdom inal region during free breathing and gating [20]... 26 2.6 Illustration of the various measured observables used to measure
respiratory-induced motion, (a) Lung tum our motion (b) surface render of the ex ternal thoracic markers on a patient, (c) The surrogate organ is the bottom of the lung and the tum our is illustrated in green... 28 2.7 Illustration of partial Cartisian traces of three-dimensional tum our res
piratory motion in four different patients as explained in [58]... 29
3.1 Schematic representation of the proposed m ethod for extracting respiratory-induced motion from the dynamic CT d ata sets (simulated or clinical d ata sets)... 39 3.2 Illustration of the helical CT image acquisition process [16]... 41 3.3 Illustration of the image scan in ciné mode and reconstruction of the
multi-slice CT (MSCT) image d ata sets as described in [90]. In this dia gram, four images are reconstructed per sample (dot) for a 4-slice MSCT and 8 samples (arrows) in each respiratory cycle (sinusoidal curve), where these images are obtained at regular intervals at each table position. For each table transition (or gantry rotation), the x-ray remains turned off. Finally, these slices are grouped into volumes for each similar phase of the respiratory signal... 42 3.4 Illustration of the anatom y of the 4D XCAT phantom and the simulated
chest x-ray CT images from the XCAT phantom. Coronal(top row), transaxial (bottom 2 rows) reconstructed slices according to [102]. . . . 42 3.5 Segmented and visualised from itksnap the (a) orignial CT d ata and (b)
the segmented tum our and organs at risk from patient 1 of P O P I d ata . 46 3.6 Segmented and visualised using itksnap of the 3D models of the seg
mented organs’ surface renders from patient 1 of P O P I d ata (a) lungs (b) liver (c) kidneys (d) spleen... 47 3.7 Illustration obtained through MATLAB, the surface render of the lesion
th a t was previously segmented from the p atien t’s dynamic CT data. . . 48 3.8 Illustration of the decomposed point cloud obtained from the surface of
the segmented lesion displayed above, as a preprocessing step for the image registration... 49 3.9 Illustration of the registered point clouds of consecutive frames of the
lesion P O P I d ata set-Patient 1, where the red and blue point sets depict two different phases of the breathing cycle of the lesion... 57
3.10 Illustration of the convergence of IC P registration for a lesion registered to the consecutive phase of the breathing cycle. The fitting error was observed over 15 iterations... 58 3.11 Illustration of the diaphragm motion in the SI direction over a period of
10 phases (one breathing cycle)... 59 3.12 Illustration of the obtained IC P error values of all segmented organs over
one breathing cycle... 60
4.1 Illustration of the use of PC A for dimensionality reduction by [125]. The maximum spread of the cluster represents the overall motion along the first principal component (the big red axis) and th e small red axis represents the second principal component and its spread respectively. 66 4.2 Illustrations of the principal components obtained when PCA was per
formed on the lung tum our after quantification of its motion. . . . 72 4.3 PCA on the external markers of VENI d ata illustrates th e highest vari
ance along the lateral direction... 73 4.4 Signal variation on the top most external surface marker when perform
ing the P C A ... 74 4.5 P lot illustrating th e correlation of the external marker (s) m otion w ith
respect to the lung tum our motion in eigenspace... 75 4.6 Velocity plot illustrating the directionality of motion w ith their respec
tive velocities... 75 4.7 Illustration of the PCA performance when considering only th e entire
right lung as the internal surrogate... 76 4.8 Illustration of the signal variation observed when considering only th e
entire right lung as th e internal surrogate... 77 4.9 Illustrates the correlation coefficient of the registered param eters be
tween the lesion and the entire right lung in eigenspace... 78 4.10 Illustration of th e velocity change observed when considering th e entire
right lung... 78 4.11 PCA performed when considering both lungs as th e internal surrogate. . 79 4.12 Signal variation observed when considering both lungs as th e internal
surrogate... 79 4.13 Illustrates the correlation coefficient of the registered param eters be
tween the lesion and both lungs in eigenspace... 80 4.14 Illustration of the velocity change observed when considering b o th lungs. 81 4.15 PCA performed from the viewpoint of the bottom of th e right lung as
xii List o f Figures
4.16 Signal variation observed when considering bottom of the right lung as a viewpoint and as th e internal surrogate... 82 4.17 Illustrates the correlation coefficient of the registered param eters be
tween the lesion and the bottom of right lung in eigenspace... 83 4.18 Illustration of the velocity change observed when considering the bottom
of the right lung... 83 4.19 PCA performed when considering the liver as the internal surrogate. . . 84 4.20 Signal variation observed when considering the liver as the internal sur
rogate... 85 4.21 CCA when considering the liver as th e surrogate... 85 4.22 Velocity change observed when considering the liver as th e internal sur
rogate... 86 5.1 Illustration of the internal surrogate in this case the lungs along with
the tum our... 92 5.2 Figures (a)-(c) illustrates the mean displacement of the RMS error of
the lung lesion (tum our), lungs and liver for P atient 1 of the P O P I data set... 93 5.3 Illustration of the fitting error and convergence graph of lungs from
P O P I /P l when registering from phase 2 (source) to phase 1 (reference). 94 5.4 Illustration of the lesion center translations dem onstrating error varia
tions along each direction (LR, AP, SI) for patient 1 of P O P I data. . . . 95 5.5 Illustration of the displacement errors of the rotation param eter for pa
tient 1 of PO PI d a ta ... 95 5.6 Illustration of the displacement errors of the scaling param eter for patient
1 of P O P I d a ta ... 96 5.7 Illustration of the displacement errors of the shear param eter for patient
1 of P O P I d a ta ... 97 5.8 Illustration of PCA performed on the lesion of P O P I P I data. Although,
12 DOF were used due to the smaller values it does not appear on the plot as PCA has made redundant of such registration param eters. . . . 98 5.9 Illustration of the signal variation th a t occurs when PCA was performed
on the extracted lesion (tum our) motion of P O P I P I d a ta ... 98 5.10 Illustration of external surface markers on patient 1 of P O P I data. . . . 100 5.11 Illustration of the motion p ath of the top most anterior marker obtained
from clinical 4D CT d ata set... 100 5.12 Illustration of the correlation of the lung tum our motion with respect
to the external surface marker motion in eigenspace. S is the tho racic / abdom inal surface marker with respect to lesion motion in LR, AP and SI directions... 101
5.13 Illustration of th e lung tum our and the bottom of the right lung marked in blue. The tum our is illustrated in green... 103 5.14 Illustration of the principal components obtained from th e lungs of pa
tient 1 of P O P I d a ta ... 104 5.15 Illustration of th e signal variation of the extracted motion for lungs dur
ing th e breathing cycle... 104 5.16 Illustration of the directionality in the x direction of th e lesion center’s
motion w ith respect to the bottom of the lung. The s ta rt position is denoted with a red 4- sign...105 5.17 Illustration of the directionality in the y direction of th e lesion center’s
motion w ith respect to th e bottom of the lung. The s ta rt position is denoted w ith a red 4- sign...106 5.18 Illustration of th e directionality in the z direction of the tum our center
w ith respect to th e bottom of th e lung. The s ta rt position is denoted w ith a red 4- sign... 106 5.19 Illustration of the PCA performed on the liver of P O P I P I d a ta ... 107 5.20 Illustration of th e signal variation on liver when PCA was performed of
P O P I P I d a ta ... 108 5.21 Correlation model of the lung tum our and internal surrogate lungs in
eigenspace... 109 5.22 Correlation model of the lung tum our and internal surrogate liver in
eigenspace... . 1 1 0 B .l Illustration of th e linear accelerator [120]... 120 B.2 MLC collimator adjusting according to th e shape of the tum our [1]. . . 121
2.1 Summary of the 3D tum our motion d a ta properties w ith dom inant mo tion shown in bold where X,Y and Z correspond to the right-left, superior- inferior and anterior-posterior orthogonal directions of motion, respec tively [58]... 27 3.1 Anatomical and respiratory motion param eters of the XCAT phantom
for a m a l e ... 43 3.2 Param eters used for generating 4DCT image d a ta sets from 4D XCAT
p h a n to m ... 58
5.1 The patient param eters for 5 patients each w ith 10 phases of breathing (or one breathing cycle)... 90
C .l Coordinates obtained from th e top most external surface m arker from clinical d a ta ... 123
In trod u ction
E xternal beam radiotherapy (EBRT) is a vastly popular m ethod in treatin g cancer patients. However, when treating cancer patients w ith radiotherapy, it is to be noted th a t the involuntary movements in a patient, could cause undesirable effects th a t would later determ ine the dosimetry. Such involuntary movements include respiratory mo tion, digestion, blood flow, e tc ... are to nam e a few. In this research however, focus is mainly on th e affects on radiotherapy procedures due to respiratory-induced motion. Respiratory-induced motion causes ambiguities th a t can distort the shape of an organ or tum our itself, th a t could lead to undesirable dose delivery in th e treatm en t stages in radiotherapy procedures for patients th a t have developed tum ours in thoracic and abdom inal regions, mainly lung cancer patients. This respiratory-induced m otion if not properly treated, could lead to higher dose delivery which could cause to lower th e levels of healthy tissue sparing or lower dose delivery which could expedite the growth of the tum our. It is therefore im portant to find a feasible motion model th a t could account for respiratory motion during radiotherapy treatm ent which would ultim ately lead to th e reduction of the planning targ et volume (PTV ) margin th a t would benefit the p atien t’s treatm ent plan. Therefore, first and foremost, a brief overview of th e cur rent radiotherapy procedures, different tum our volumes, the current state of scientific understanding, technical methodology in imaging and finally, treatm ent planning and treatm ent delivery is discussed.
Chapter 1. Introduction
1.1
Thoracic and A bdom inal Cancer
Male Fem ale
Lung Bov/el’ Breast || Prostate * Pan creas O esop tiagu s Stom ach B la d d e r Leukaemia Non-Hodgkin Lymphoma Ovary Kidney Brain and Centra! Nervous System
Liver H f Other Digestive Organs j j |||| Myeloma H P M esotheliom a P * Malignant M elanoma H P O r a lJ T Uterus Other Sites** I 0 5,0 0 0 10,000 N um ber o f D ea th s 15,000 20 ,000
F ig u re 1.1: Cancer statistics for the year 2010 as issued by the cancer research UK [122]
Lung cancer is one of the most prevalent causes of death in the world currently. Ac cording to the statistics shown in Figure 1.1 as of 2010 in UK [122], for both men and women the most common cause of cancer death is lung cancer. It estimates a stag gering 19,410 deaths for men and 15,449 deaths for women in the year of 2010 in UK w ith an extremely poor survival rate of 5-10% for five years of survival. Moreover, for patients in the age range of 15-39 have the highest average of survival rate for lung cancer w ith only 26% [122]. Smoking tobacco is the single greatest risk factor for lung cancer related deaths, which accounts for 88% male and 84% female deaths in UK. O ther known risk factors th a t cause lung cancer include passive smoke and occupa tional and environm ental exposures to a carcinogen (such as asbestos, heavy metals, radon gas, talc dust, etc..) [122]. The most common cancer in the abdomen is bowel cancer with an average of 8600 for men and 7308 for women [122]. Although, the main
reason for abdom inal cancer is unknown and is still a t large, it is assumed th a t it is mainly hereditary and a person’s diet may also lead to the development of abdom inal cancer. The available treatm ent options for various stages of these cancers can be di vided into 3 main categories, namely, surgery, radiotherapy an d /o r chemotherapy. For the purposes of this research however, the focus will only be on radiotherapy treatm ent adm inistered externally to the patients during the treatm ent stages^ and in specific for tum ours in the thoracic and abdom inal cavities.
1.2
D iagn ostic Im aging
W ith the rapid development and proliferation of medical imaging technologies, scien tists and physicians are able to glean potentially life threatening inform ation by peering non-invasively into th e hum an body. Beyond the simple visualization and inspection of the anatomical structures, medical imaging has become a tool for surgical planning, simulation, intra-operative navigation, radiotherapy treatm ent planning and tracking the progress of disease [7]. These technologies include X-ray, com puted tom ography (CT), magnetic resonance imaging (MRI) and nuclear medicine imaging^ technologies. These can be divided into 2 main categories, namely, modalities th a t use ionising ra diation and modalities th a t use non-ionising radiation. W hen radiation is ionised, it releases an electron from an atom or a molecule due to its particles’ immense kinetic energy. Although, CT scans generate fast, detailed 3D volum etric image datasets, th e main disadvantage is th a t these powerful rays th a t are em itted during a CT scan, are capable of directly or indirectly damaging the DNA. Modalities such as CT and P E T use ionisation radiation when imaging. A typical CT scanner uses an X-ray source as an em itter and a detector on opposing sides as shown in Figure 1.2.
A typical CT scan generates a 3-dimensional (3D) volume of d ata obtained by im aging a large number of 2-dimensional (2D) slices around a single axis of rotation.
^ which includes treatment planning and delivery stages which will be discussed in detail later in the chapter
^which generally includes single-photon emission computed tomography (SPECT) and positron emission tomography(PET) imaging technologies.
Chapter 1. Introduction x-ray Source Nlotortï®? Table "©tactot*
F ig u re 1.2; Illustration of the operation of a CT scanner
1.3
R adiotherapy
Radiotherapy, also known as radiation therapy or radiation oncology, is a widely used localised non-invasive treatm ent m ethod [20] for cancer patients, among other invasive treatm ent m ethods such as, chemotherapy and surgery. The mechanism of the radio therapy procedure is simply to cause irreparable damage to the DNA of the cancerous cells by exposing the tum our cells to ionizing radiation. Radiotherapy can be broadly sub-categorized to three distinct radiotherapy delivery approaches, namely internal (brachytherapy), external beam radiotherapy (EBRT/X RT) and system atic radioiso tope therapy (unsealed source radiotherapy).
1 .3 .1 I n te r n a l R a d io t h e r a p y
Internal radiotherapy also known as sealed source therapy, im plant therapy, intersti tial radiation or intra-cavity radiation, is a minimally invasive procedure th a t uses radioactive sources such as radioactive iodine (I^^^), cesium (Cs^^^), iridium (Ir^®^), phosphorus (P^^), cobalt (Co^®) in implants or seeds, placed directly into or near the tum our 1.3 [83].
Brachytherapy is performed in three straight forward steps:
e s to ïe * w « » t ) « n ( O f jN i!KB!Eiiawt< <miial«M»»(c*i. j » (Opcs h w ih w.
F ig u r e 1.3: Interstitial Brachytherapy procedure performing the seed im plantation on a prostate cancer patient, using a needle [83].
clinical exam ination is perform ed to obtain a C T or M R I image scan o f the tu m o u r and to plan the optim al distribution o f treatm ent. A t this stage, the applicators or source carriers that deliver the radiation, typically, non radioactive needles or plastic catheters, will also be discussed.
In sertio n o f applicators:
In order to ensure the applicators are correctly positioned aligning with the tu m our, by using imaging technologies such as X-ray, fluoroscopy, or C T to guide placing the radioactive sources.
3D V irtual P a tie n t C reation and O ptim izin g the Irra d ia tio n Plan:
W ith the C T image scans obtained, a 3-dim ensional (3D) rendering o f the appli cators, the treatm ent volum e and the surrounding healthy tissues bear a spatial re lationship that would help optim ize the irradiation plan in the spatial and tem poral dom ains fo r the tum our. The graphical representation will guide the oncologists to deliver a patient specific dose also som etim es referred to as dose-painting.
The main advantage in using internal radiotherapy is th a t it increases th e chance of healthy tissue sparing as the radiation delivered is concentrated on or near th e tum our. One of the main disadvantages of brachytherapy however is th a t, it is unsuitable for aggressive tum ours as the radiation strength is insufficient for treatm ent of tum ours
Chapter 1. Introduction
of such magnitude. Another disadvantage is the patient will remain radioactive for several days after the treatm ent session. Internal radiotherapy is generally used in treating tum ours of the head, neck, thyroid, female reproductive system, prostate or breast.
1 .3 .2 S y s t e m a t ic R a d io is o t o p e T h e r a p y
Unsealed source radiotherapy also known as systematic radioisotope therapy or source radioisotope therapy, uses soluble radioactive substances, such as radioactive iodine th a t are administered to the body by means of infusion or oral ingestion [122]. However, one obvious disadvantage is th a t patients who received treatm ent are consid ered to be radioactive throughout the duration of the procedure and will remain so for several weeks as well much like in brachytherapy [122]. Unsealed source radiotherapy is generally used to treat patients w ith thyroid cancers.
1 .3 .3 E x t e r n a l B e a m R a d io t h e r a p y
E xternal beam radiotherapy conventionally uses photon^ beams such as, high energy X-rays, gam ma rays or charged particles to treat cancer patients. Photons allow in teractions at longer distances, therefore, they can even penetrate to trea t deep seated tum ours. E xternal beam radiotherapy procedures are carried out with the aid of linear accelerators also known as LIN AC machines [89, 120] [Appendix A describes the full functionality of a Linac machine]. Figure 1.4 illustrates external thoracic radiation therapy given to a patient suffering from pleural mesothelioma cancer.
There are numerous advantages of using EBRT as opposed to the other types of radio therapy treatm ent procedures. These mainly depend on a p atien t’s age, health and the stage of the cancer. Since, EBRT is a non-invasive procedure patients are less prone to having any of the risks of surgery. EBRT is painless and does not have the residual contam ination issues with radioactivity at the end of the session th a t is associated with
^Photons are quantum energy packets that carry no electrical charge or rest mass, but can have varying energy levels in the electromagnetic spectrum that, can ultimately emit energy rays such as, high energy X-rays, gamma rays, etc . . .
F igure 1.4; Illustration of the external beam radiation therapy performed on a patient with pleural mesothelioma cancer (cancer attacking the membranes of the heart, lungs and the abdom inal cavity) [129].
unsealed source therapy. Unlike in brachytherapy, patients will also experience less discomfort during treatm ent sessions and will not be contam inated. E xternal beam radiotherapy is the most widely used m ethod of cancer treatm ent and is generally given to all types of cancer patients depending on the type and stage of cancer. EBRT can be delivered in many forms, namely, conventional, stereotactic radiation, particle therapy and advanced imaging technologies using 3D-virtual simulation for treatm en t (3-dimensional conformai radiotherapy (3D-CRT), intensity m odulated radiotherapy (IMRT), image-guided radiotherapy (IGRT)). For the purposes of this research the focus is confined to EBRT treatm ent methods.
1 .3 .4 T r e a tm e n t P la n n in g a n d T r e a t m e n t D e liv e r y
The radiotherapy treatm ent process prim arily consists of two stages, namely, treatm ent planning and treatm ent delivery. The core objective in the treatm ent planning phase is to find treatm ent param eters th a t will maximise the treatm ent effect in the targ et region while reducing potential complications in the normal tissue surrounding the tum our. In the treatm ent planning phase, typically, a com puted tom ographic (CT) or magnetic resonance imaging (MRI) scan over the region of the tum our, is acquired for the radiation oncologist to determine the stage of the tum our and plan a treatm ent
Chapter 1. Introduction
delivery dose over the next few weeks (also known as fractionated treatm ent). The first step is to delineate the volume margins around the target tum our volume. These volume margins contain the tissue volumes th a t will potentially be irradiated to a specified dose. These volume margins usually account for some level of expected motion, expected variations in shape and size and also the variations in treatm ent set up, and therefore, need to be large enough to ensure coverage of the target [20].
D e lin e a tio n o f T um our M argins
A radiation oncologist will delineate the palpable or visible/dem onstrable extent of the malignant growth of the targeted tum our on the acquired CT image, called the gross tum our volume (CTV) [2, 20, 84, 87, 88]. Thereafter, accounting for any suspected microscopic spread the clinical target volume (CTV) margin is obtained [2, 84, 87, 88]. Finally, CTV margin is extended to account for geometric errors such as averaging or blurriness, due to intra-fraction motion^, inter-fraction motion^ and set up errors, the planning target volume (PTV) margin is delineated, to which the dose is finally pre scribed [2, 84, 88]. Since, many advances in medical imaging technologies have emerged during the last few years, target volume positions are now easier to measure, their ac curacy increased and measurements are taken more frequently for individual patients as opposed to previous m ethods of determining the margins, as it required according to [87], to accommodate a population of patients in order to estim ate a margin. There fore, the inter-fraction errors, set-up errors will be negligible and would just take into account the geometric errors (intra-fraction, measurement and correction errors) [87]. The volume th a t encompasses the aforementioned margins (isodose surface) represent ing the planned minimal target dose is known as the treated volume (TV). The tissue volume th a t receives the dose in reality is considered to be the irradiated volume (IrV) [87]. The treated volume along with the CTV, CTV, PTV margins and the irradiated tissue volume is illustrated in Figure 1.5.
^occurs within a treatment session such as respiration ^occurs between treatment sessions
'Zx
Ms#
WM
k
'■ ;(b)
F igu re 1.5: Schematic illustrating the process of treatm ent planning: (a) CT image illustrating the tum our on th e right lung and is used as reference to delineate the target volume margins, (b) Zoomed in on the area where target volume margins are delineated to show the C TV (red line), extending for microscopic spread CTV (green line) margin, and finally PTV (yellow line) accounting for intra-fraction errors and to which the dose is prescribed. Treated volume margin (purple line) and finally th e irradiated volume margin in (blue line) according to the ICRU report 50 [84]. [Note: CT image th a t was used to delineate was obtained from patient 1 of P O P I [123] d a ta set].
1 .3 .5 L im it a t io n s
Involuntary internal motion such as respiratory motion is one of th e prim ary causes of tum our motion in the thorax and abdomen, which creates lim itations in the treatm ent planning phase. Current practices require patients to perform shallow breathing or free breathing during the scan process (treatm ent planning) and treatm ent delivery phases. This gives rise to lim itations in image acquisition, treatm ent planning and treatm ent delivery phases. In conventional radiotherapy procedures, internal m otion (respiration) if not accounted for, will cause artifacts th a t would distort th e target volume and give incorrect positional and volumetric inform ation as it will be difficult to delineate the boundaries of small mobile volumes th a t are potentially cancerous [20]. In the treatm ent planning phase, it is im portant to account for tum our motion due to respiration, therefore, the volume margins need to be sufficiently large. Figure 1.6
10 Chapter 1. Introduction
illustrates the full motion p ath of the tum our during a single breathing cycle. Therefore, in reality the entire tum our motion limits need to be taken into account in order to predict th e exact dosimetry for the patient. This may lead to increased radiation field size and consequently increase in the volume of healthy tissue exposed to the radiation. Finally, in th e treatm ent delivery phase, an averaging or blurring of the static dose distribution over the path of motion will occur if respiratory motion is not accounted for [20]. Consequently, there will be a displacement between the intended and delivered dose distributions (also known as hysteresis). Therefore, it is of utm ost im portance to account for any limitations at any stage of the treatm ent procedure.
F R E E B R E A T H IN G
F ig u re 1.6: Illustration of the full motion p ath of the tum our and its margins due to respiration.
R adiotherapy treatm ents when given both externally and internally to the patients, can be classed as therapeutic (curative), prophylactic and palliative radiotherapy. Ther apeutic (radical) radiotherapy focuses on destroying the tum our. Prophylactic radio therapy is to stop the cancer growing in the same area after surgery and to reduce metastasising to other organs. On the other hand palliative radiotherapy is used for lessening pain or reducing the size of the tum our. Palliative treatm ent is usually of lower doses th a n therapeutic radiotherapy and is given over a shorter period of time. Typically, the am ount of radiation dose delivered is measured in gray units (Gy) and varies between the type and stage of the cancer being treated. For therapeutic cases such as lym phoma are treated typically with a dose of 20 to 40Gy, while for epithelial tum our the dose would be in the range of 60 to 80Gy [121, 123]. Delivery param eters for a prescribed dose are determined during the treatm ent planning phase. Depending
on th e stage and type of cancer these prescribed doses can be spread out over a period of tim e (or fractionated).
1.4
S tate-of-th e-A rt Im aging and G uidance
Among much advancements in radiotherapy two of the salient advancements are, first, the availability of imaging in the treatm ent room, allowing improvement to th e delivered treatm ent accuracy of ±3m m . Second is th e development of respiratory correlated imaging, providing patient specific respiratory motion for treatm ent planning, among which the leading image techniques are 4-dimensional com puted tom ography (4DCT) and cone-beam com puted tom ography (CBCT).
One of the earliest, most common treatm ent m ethods is th e conventional radiotherapy also known as 2-dimensional external beam radiotherapy (2DXRT), A single beam of radiation is delivered in difierent directions to the patient. The main advantage of using conventional m ethod is th a t since the radiation delivery is planned or simulated on a specially calibrated diagnostic x-ray machine known as a simulator because it recreates the linear accelerator actions in order to accurately target and localize, it is considered to be quick and reliable. However, the m ain disadvantage of this m ethod is th a t th e actual treatm ent volume margin is much greater th a n th e target tum our volume itself. Therefore, volume of radiation exposure of norm al healthy tissues surrounding th e tum our is greater.
Proton therapy is another interesting m ethod of EBRT, where energetic ionizing p a rti cles such as protons or carbon ions are directed at target tum our. The radiation dose increases while th e particle penetrates the tissue reaching th e Bragg peak which is th e maximum limit th a t occurs near the end of the particle’s range and drops to alm ost a zero value [20]. The main advantage in this m ethod is th e am ount of norm al healthy tissue exposure to less radiation.
12 Chapter 1. Introduction
1 .4 .1 V ir t u a l s im u la tio n
3D -C on form aI R a d io th era p y
One of the most common external beam radiotherapy methods is the three dimensional conformai radiotherapy (3D-CRT). 3D-CRT uses highly innovative com puter software and advanced treatm ent machines to deliver radiation to very precisely shaped target areas [121]. 3D-CRT uses a multi-leaf collimator (MLC) to conform the shape of the tum our and deliver the radiation dose. Thus, this computer-guided m ethod delivers high-dose radiation with relatively fewer side effects.
In te n s ity M o d u la ted R a d io th era p y
W ith the recent advances medical imaging technologies, a widespread technique is the high-resolution intensity-m odulated radiation therapy (IMRT), which is sometimes called three-dimensional IMRT (3D-1MRT), and uses a multi-leaf collimator (MLC) as well. During this treatm ent, the patient breathes normally while receiving a 4DCT gated acquisition where, the layers of the multi-leaf collimator are moved while the treatm ent is being given. This m ethod is able to shape the profile of the treatm ent beam s very precisely and allows the dose of external radiotherapy to be altered in different parts of the treatm ent volume. The main goal in IMRT is to increase radiation dose to the tum our volumes while reducing the radiation exposure to specific sensitive volumes surrounding the normal tissue. IMRT has fewer side effects when compared to traditional external 3D conformai radiotherapy (3D-CRT) treatm ent, such as damage to the salivary glands which can cause dry m outh when head and neck are treated with IMRT [121].
Im a g e G u id ed R a d io th era p y (IG R T )
Image-guided radiotherapy is a widespread radiotherapy treatm ent m ethod as it pro vides the tools needed to manage both inter- and intra-fraction motion to improve th e accuracy of radiation field placement during treatm ent delivery [50]. IGRT uses a technique called ’’gating” which is an electronic m ethod of limiting the effect of normal
intra-fraction motion. G ating enables the radiation beam to selectively tre a t a moving tum our by turning the beam on/off at specified intervals [72, 93, 94]. One of th e salient features of IGRT is th a t it takes into account of all th e possible set up errors, patient immobilization, e t c ... during treatm ent.
1.5
M otivation
Motion between image frames (CT or P E T or MRI) during one radiotherapy session, also known as intra-fraction motion has become a crucial issue in image guided radio therapy. Intra-fraction motion can occur due to motion in respiratory, skeletal mus cular, cardiac and gastrointestinal systems [20]. However, this research will focus on intra-fraction motion caused due to th e respiratory system as it is one of the most common problems occurring in external beam radiotherapy procedures. The tum our m otion induced by patients free breathing during treatm ent delivery limits the accu racy of radiation dose delivered to the target. This will lead to system atic dose delivery errors, damaging not only th e cancerous cells b ut also th e healthy cells surrounding th e tum our in a patient, as opposed to 3-dimensional conformai radiotherapy (3D-CRT), where th e beam is shaped to conform to the shape of the target while limiting th e dose delivered to th e organs at risk. In order to achieve th e goal of norm al tissue sparing, a patient needs to lie still while the treatm ent is in progress. However, if any patient movement is detected it could cause undesirable ambiguities in th e location of th e tu mour. In contrast to previous literature, where assum ption is m ade th a t th e external markers and thoracic/abdom inal surface moves in unison w ith the tum our motion, this research will study the param eters of the p atien t’s actual breathing motion and will build a param eterised spatio/tem poral motion model. A nother common problem is the radiation beam delay or latency due to system [85]. The delay in the tim e needed for the radiotherapy machine to move the head and produce a beam profile th a t conforms to the treatm ent plan at a particular angle. This means th a t th e p atien t’s internal organ configuration needs to be dynamically predicted in order to minimize treating normal tissue at the tum our margin [58].
14 Chapter 1. Introduction
Organ motion will result in a deviation of the delivered dose from treatm ent plan. If not accounted for, such organ motion due to respiration, can lead to an under-dose of tum our or overdose of the surrounding tissue [61]. By defining a patient specific motion model, the organ motion of the anatom y of interest due to respiration will be described, and in a more general context is assumed th a t the motion model is capable of facilitating th e respiratory-induced motion. Thus, accounting for motion will improve the effectiveness and efficiency of the radiotherapy treatm ent.
1.6
P rop osed M eth od
In this research therefore, emphasizes in reducing the planning target volume (PTV) margin for radiotherapy procedures. Figure 1.7 illustrates the grand schematic for the overall problem of reducing the PT V margin. First stages of treatm ent planning will be, to obtain dynamic CT (upto ten volumes or one full breathing cycle), which is initially obtained through a digital phantom called XCAT, (which will be explained in detail in a later chapter) th a t can be made to represent the organ motion as it appears in CT in a rather idealised manner. This m ethod will later use dynamic CT of real patient d ata to obtain a valid motion model. The motion model is perturbed for changes in breathing cycle accordingly, as individual characteristics of breathing (quiet/shallow vs. deep, chest vs. abdominal, etc..) and motion variations associated with tum our location and pathology could lead to distinct individual p atterns in displacement, direction and phase of tum our motion. This is achieved by using iterative registration m ethods [10] to determ ine the transform ation param eters in order to localise the tum our and various organs’ motion (OAR) at different phases, and from this infer tum our motion. It is im portant to note the following were assumed in this case;
• the organs behave in a piecewise affine motion, • th a t no movement occurs due to digestion, • the breathing is consistent throughout the scan.
Therefore, w ith the aid of registration algorithms a param etrised tum our volume could be obtained and the tum our position could be determined. This m ethod is also
concomi-ta n t in finding the correlation of motion between th e lung tum our and internal/ external surrogates. The correlation model between th e lung tum our and th e surrogates allow to observe th a t the tum our volume could be targeted within ± 3 m m from th e CTV - P T V m argin in any direction which accounts for geometric errors in treatm en t planning. Consequently, reduces the overall P T V margin of error in tum our m otion considering imperceptible penetration, w ith a standard deviation of ±1%, which leads to reduced patient dose delivery in the treatm ent plan, where significant norm al tissue sparing could be gained.
1.7
O utline o f T hesis
The first chapter discusses briefly about th e radiotherapy procedures, m ethods th a t are currently in use and also the margins and th e treatm ent plans and their m ethods of delivery. It also details the state-of-the-art imaging technologies th a t are being used to trea t patients. The second chapter discusses in depth the physiology and the breathing mechanics to help understand the different patient models. It also reviews the different respiratory motion management m ethods th a t are currently in use.
The th ird chapter details the key points of extracting the respiratory-induced m otion from four dimensional com puted tomographic (4DCT) data. The iterative registration m ethods in specific, IC P algorithm was used to help extract this motion. This d a ta was both in the form of simulated and clinical d a ta sets. The fourth chapter discusses respiratory-induced motion correlation model. It details the param eters th a t were used to build the correlation model and how it will be implemented. This also examines th e preliminary results of the correlation model. Next chapter implem ents th e proposed respiratory correlation model on the clinical d ata sets th a t was obtained via th e CRE- ATIS labs in France. This analyses the feasibility of the proposed m ethod. Finally, th e last chapter briefly discusses the outcomes and possible future work.
1.8
A chievem ents
16 Chapter 1. Introduction 4^ of PTV m argin In EBRT U p d a t e d o s i m e t r y T re a tm e n t Delivery T re a tm e n t Planning C a l c u l a t e c h a n g e s s i n c e p la n n i n g T u m o u r l o c a l i s a t i o n u sin g d y n a m i c CT C o m p e n s a t e for m o t i o n an d c o rrect PTV m argin Build Correla ti on M o t i o n M o d e l
F ig u re 1.7: Schematic representation of the proposed method. A dynamic CT volume will be used to param eterise the tum our location and subsequently tum our motion is observed in order to determine the tum our motion pathway with respect to the p atien t’s breathing pattern. This is then used to determine the entire tum our motion p ath and subsequently reduce the PT V margin for dose delivery.
R espiratory-induced Tum our
M otion and M otion M od ellin g in
R adiotherapy
Imaging is th e first step in the treatm ent stages of radiotherapy. W hen imaging (CT or MRI), distortions occur due to involuntary movements such as breathing and will subsequently lead to erroneous dosimetry th a t could be undesirable to th e patient. Specifically, in the planning stage where delineation of targ et tum our volumes and the localisation of adjacent normal structures th a t determ ine the patient dosim etry are carried out. This motion however, as in all cases will depend greatly on th e location of the tum our and the region th a t the given motion is affected. Respiratory-induced in tra fraction motion ^ is a budding problem in the treatm en t planning and delivery stages in radiotherapy procedures. Initially, the treatm ent planning and delivery were developed as static processes, in which th e intended treatm ent plan would be delivered only if the tum our was stationary at th e beam isocenter. Thus, internal motion in patients causes inaccurately aiming th e radiation beam to the targeted lesion, leading to erroneous detection and tracking of thoracic malignancies (tum ours) in the treatm ent planning stage. This positional uncertainty due to respiration is inherent in radiotherapy, and
^Intra-fraction motion is motion that occurs within one session. Inter-fraction m otion is m otion that occurs between sessions.
18 Chapter 2. Respiratory-induced Tumour M otion and M otion Modelling in Radiotherapy
the prevailing method according to the ICRU report 50 [84] is to irradiate a volume larger th an the target. Consequently, the radiation dosimetry for the tum our in the treatm ent delivery stage can potentially be harmful to the patients as it will harm any healthy tissues surrounding the tum our. Therefore, it is im portant to account for any internal target motion th a t is induced by respiration, th a t occurs during treatm ent planning and delivery.
This chapter will look at the physiology of respiration, respiratory-induced motion in radiotherapy, the specific problems th a t are arisen due to respiratory-induced motion and techniques th a t account for the aforementioned respiratory-induced motion.
2.1
P h ysiology o f th e R espiratory System
Anatomically, the respiratory system can be classified according to structure and fu n c
tion. Structurally, the respiratory system consists of the upper respiratory system th a t
includes, nose, pharynx and associated structures and lower respiratory system th a t in cludes the larynx, trachea, bronchi and lungs [119]. The respiratory system ’s function can also be further subdivided into two categories, namely, the conducting zone which consists of interconnected cavities and tubes th a t filter, warm and moisten air th a t will be conducted to the lungs later and the respiratory zone consisting of lung tissue responsible for the gas exchange (explained in greater detail later in the chapter).
2.2
M echanics o f R espiration
Respiratory motion is, an involuntary internal motion th a t allows the lungs to facil itate its cardinal function of the exchange of oxygen (O2) and carbon dioxide (CO2) between blood and air in th e body. As illustrated in Figure 2.1 gas exchange occurs in the respiratory zone due to diffusion between the alveoli and the blood in lungs’ capillaries [119, 128, 131]. W hen the dissolved gases have entered the blood stream the heart will carry the essential gases to the vital organs via the circulatory system. Al though, respiratory motion is involuntary and is not rhythm ic as the heart, frequency
Nbsw cwity Alveoli Ep^oltis Lwynx rreshea . C«rt>on Left \ daxtde (C O ^
\ Oxygen (Oj) (croM* Alveoli KN I Pulmonery lunas I * * « y (Mood from h ea rt) — «
o
wrsp3cc COs0m ' Caplltarieft * Alveolar C ep ae ry oe&vofk lurreunding Cube* (prv^oys) WHhm lyng Rib! Pulm onary ttieoOto heart)F igu re 2.1: Illustration of the anterior view of the structure of th e respiratory system [53]
and m agnitude of the respiration can be controlled within limits, as well as breath- holds [20]. R espiratory motion p atterns are categorized according to posture (upright, prone, supine, lateral decubitus), breathing types (chest or abdominal) and depth of respiration (shallow , normal, deep)^ [20, 53, 112, 119].
Figure 2.2 illustrates the typical breathing mechanics of the lungs during a normal respiratory cycle.
2 .2 .1 I n h a la tio n
Lungs are held in the thoracic cavity encased by the liquid filled intra-pleural space [20]. During quiet breathing^, respiratory muscles around the lungs actively participate in the process of breathing. It is however to be noted, th a t the principle muscle during inhalation is the diaphragm and is dome-shaped, concave towards the abdom en [128]. O ther respiratory muscles th a t connect the adjacent ribs such as external intercostal
^Normal pattern of quiet breathing is called eupnea. Eupnea can be shallow, deep or combined shallow and deep breathing. Shallow breathing is also known as chest breathing, normal breathing is also known as costal/tidal breathing or normal resting breathing and deep breathing is also known as diaphragmatic breathing [119]
^Due to the fact that majority of the air is drawn into the lungs during quiet breathing is considered as opposed to other types of breathing patterns.
20 Chapter 2. Respiratory-induced Tumour M otion and Motion Modelling in Radiotherapy
breathing In /
chest expands i Rbreathing out
ribs diaphragm diaphragm contracts© 2006 Encyclopedia Britannica, Inc.
chest contracts
tang
diaphragm relaxes
F ig u r e 2.2: Physiology of typical healthy lungs during inhalation [29].
muscles, slope downward (interiorly) and forward (anteriorly). During inhalation the volume of the thoracic cavity increases and air is drawn into the lungs. As illustrated in Figure 2.2, th e volume of the thoracic cavity increases partly due to the contraction of the diaphragm , which causes the diaphragm to push downwards (interiorly) causing the abdom inal organs to descend as well as move forward (anteriorly), and partly due to the external intercostal muscles th a t raises the ribs upwards (superiorly) and anteriorly, thus increasing the cross-sectional area of the thorax both in the lateral and anterior- posterior (AP) direction. Subsequently, this reduces the intra-thoracic pressure th a t allows air to be drawn into the lungs through diffusion [20, 128].
2 .2 .2 E x h a la t io n
During quiet breathing expiration, it is considered to be passive as it is considered to be at equilibrium or at the most relaxed state. Lungs are generally elastic, therefore, returns passively to its preinspiratory volume during resting breathing and is easily dis tended. The principle muscles during exhalation are those of the abdom inal wall which include the rectus abdominus, internal and external oblique muscles and transversus abdominus [128]. During exhalation, as illustrated in Figure 2.2 the volume of the
thoracic cavity decreases w ith the aid of internal intercostal muscles pulling the ribs interiorly and inwardly (posteriorly). The intercostal muscles tighten as well in order to avoid the ribs from bulging outwards. O ther external factors such as controlled breath ing in yoga, exercise, e tc ... will have considerable effect in increasing the exhalation process.
However, it is to be noted the inflated volume of the lung a t inhalation is considerably smaller th an the deflated volume o f the lung at exhalation'^ during a breathing cycle. This asym m etry in respiration is well-known as hysteresis.
2 .2 .3 H y s t e r e s is
Hysteresis in respiration is a common occurrence in th e lungs due to the aforementioned asymmetric nature of a breathing cycle. Typically, two simultaneous projection images at oblique angles are required to locate a tum our in 3-dimensions. However, the tum our may not always follow the same trajecto ry at all times [107]. This phenom ena is called hysteresis. Realistically, occurrence of hysteresis is to be expected and is calculated the maximum distance as the inhalation and exhalation trajectories. Although, different metrics have been proposed for measuring th e maximum distance such as Hausdorff distance, Prechet distance in m ajority of th e cases the maximum euclidean distance has been considered the basis for calculating hysteresis.
Previous literature discusses th e quantification of hysteresis and the non-linearity of respiratory motion when estim ating or predicting or tracking respiratory m otion or building respiratory motion models [13, 20].
As illustrated in Figure 2.3, the shape of the trajectories curve remains fairly constant in time. Previous literature also discusses th a t the average tum our position is the most stable and reproducible during exhale phase of tidal breathing [107, 112].
22 Chapter 2. Respiratory-induced Tumour M otion and M otion Modelling in Radiotherapy
Lung-Chest Wall Pressure-Volume Curve 3.5 Deflate % 3 3.0 Inflate 2.5^ DISTENDING PR ESS U R E ( c m H f )
F igu re 2.3: Trajectories of the hysteresis curve
2.3
R espiratory M otion in R adiotherapy
As previously explained, imaging as the first step in the treatm ent process of radiother apy. Imaging technologies such as CT and MRI studies provide geometric information in the treatm ent planning phase to delineate the gross tum our volume (GTV) m ar gins and to localise any adjacent organs-at-risk (OAR). This obtained geometric data will then be used in the planning process to orient the beams to encompass the target volume while sparing any normal tissues surrounding the tum our. In CT particularly inform ation on tissue densities, th a t help calculate the dosimetry for radiation, will also be provided. Chen and Reitzel [21], explains th a t unlike in diagnostic scans, in treatm ent planning scans in th e thorax and the abdomen are often performed with the patient breathing ’’lightly” . This is performed so th a t the patient breathing is kept consistent during the scanning and treatm ent delivery. Organs in the thoracic-abdomen region move almost periodically along the cranio-caudal^ and anterior-posterior (AP) directions during respiration. In [56], describes the am plitude of this motion is on the order of centimeters with a period of ~4s.
It is im portant to note th a t in the treatm ent stages of radiotherapy, a static anatomy is assumed when delineating the tum our margins and the intended plan for the patient dosimetry would be delivered only if the tum our is stationary at the beam isocenter.
Treatm ent sites can move under a variety of physiological influences, both stochastic and determ inistic (i.e., regular, system atic and periodic) [85]. This motion would gen erally lead to erroneous expansion of the treatm ent margins and dose deliveries th a t encompass the range of motion due to respiratory-induced intra-fraction targ et (tu mour) motion (during a single fraction of treatm ent) or inter-fraction motion (different fractions over the treatm ent plan).
Although, m any studies [40, 41, 92, 106, 107, 114] have been carried out for determ ining the m agnitude and type of variation in intra-fraction variation in respiratory motion the results of these studies have proven inconclusive. One m ajor problem according to [80], is the difficulty in imaging the internal respiratory motion for th e tim e taken to deliver a fraction of treatm ent several minutes later. Currently well-designed margins are considered to deliver intended high radiation doses to th e target, whilst destroying healthy tissues surrounding the tum our. Thus, inaccurate projection of th e range of intra-fraction motion of th e target will result in any inaccurate dose delivery. Therefore, in order to b etter deliver efficient radiation doses to patients, well-designed margins need to be adaptable and possibly compensate any target motion during treatm en t sessions. To adapt and compensate, for any form of motion in th e target, in this case the lung tum our, will take time. Therefore, in order to account for any tim e delays between determ ining the tum our position or tum our localization and th e response tim e would require tum our motion prediction.
Figure 2.4 illustrates th e key components typically to consider when com pensating for the aforementioned target motion, particularly, respiratory-induced intra-fraction motion. In the motion estim ation phase, given any breathing p attern , estim ation of th e current location of th e tum our in the thorax or tum our localization is carried out. Typically, image registration m ethods and regression m ethods are used for this purpose. Once the current tum our position is localized then this inform ation is used to predict the tu m ou r’s next feasible position in the breathing cycle. Thus, w ith the predicted position a correlation model can be built, w ith the aid of external markers a n d /o r using a surrogate organ, to compensate th e targ et motion. Thus, planning volume target (PTV ) margin can be reduced and consequently patient dose will be reduced as well as normal tissue sparing will also be achieved.
