The segmentation problem arises in the elaboration of a radiation therapy plan. After the cancer has been diagnosed and the radiation therapy sessions have been prescribed, the physician has to locate the tumor as well as the organs situated in the radiation field, called the organs at risk. The physician also has to determine the different dosage he wants to deliver in each of them and has to define a lower bound on the dosage for the tumor (which represents the minimum amount of radiation that is needed to have a sufficient control of the tumor) and an upper bound for each organ at risk (which represents the maximum amount of radiation that an organ can receive without damaging). Designing a radiation therapy plan that respects these different bounds of dosage is a complex optimization problem that is usually tackled in three steps. The segmentation problem is one of them.<p><p>Mathematically, the segmentation problem amounts to decomposing a given nonnegative integer matrix A into a nonnegative integer linear combination of some binary matrices. These matrices have to respect the consecutive ones property. In clinical applications several constraints may arise that reduce the set of binary matrices which respect the consecutive ones property that we can use. We study some of them, as the interleaf distance constraint, the interleaf motion constraint, the tongue-and-groove constraint and the minimum separation constraint.<p><p>We consider here different versions of the segmentation problem with different objective functions. Hence we deal with the beam-on time problem in order to minimize the total time during which the patient is irradiated. We study this problem under the interleaf distance and the interleaf motion constraints. We consider as well this last problem under the tongue-and-groove constraint in the binary case. We also take into account the cardinality and the lex-min problem. Finally, we present some results for the approximation problem. <p><p>/Le problème de segmentation intervient lors de l'élaboration d'un plan de radiothérapie. Après que le médecin ait localisé la tumeur ainsi que les organes se situant à proximité de celle-ci, il doit aussi déterminer les différents dosages qui devront être délivrés. Il détermine alors une borne inférieure sur le dosage que doit recevoir la tumeur afin d'en avoir un contrôle satisfaisant, et des bornes supérieures sur les dosages des différents organes situés dans le champ. Afin de respecter au mieux ces bornes, le plan de radiothérapie doit être préparé de manière minutieuse. Nous nous intéressons à l'une des étapes à réaliser lors de la détermination de ce plan: l'étape de segmentation.<p><p>Mathématiquement, cette étape consiste à décomposer une matrice entière et positive donnée en une combinaison positive entière linéaire de certaines matrices binaires. Ces matrices binaires doivent satisfaire la contrainte des uns consécutifs (cette contrainte impose que les uns de ces matrices soient regroupés en un seul bloc sur chaque ligne). Dans les applications cliniques, certaines contraintes supplémentaires peuvent restreindre l'ensemble des matrices binaires ayant les uns consécutifs (matrices 1C) que l'on peut utiliser. Nous en avons étudié certaines d'entre elles comme celle de la contrainte de chariots, la contrainte d'interdiciton de chevauchements, la contrainte tongue-and-groove et la contrainte de séparation minimum.<p><p>Le premier problème auquel nous nous intéressons est de trouver une décomposition de la matrice donnée qui minimise la somme des coefficients des matrices binaires. Nous avons développé des algorithmes polynomiaux qui résolvent ce problème sous la contrainte de chariots et/ou la contrainte d'interdiction de chevauchements. De plus, nous avons pu déterminer que, si la matrice donnée est une matrice binaire, on peut trouver en temps polynomial une telle décomposition sous la contrainte tongue-and-groove.<p><p>Afin de diminuer le temps de la séance de radiothérapie, il peut être désirable de minimiser le nombre de matrices 1C utilisées dans la décomposition (en ayant pris soin de préalablement minimiser la somme des coefficients ou non). Nous faisons une étude de ce problème dans différents cas particuliers (la matrice donnée n'est constituée que d'une colonne, ou d'une ligne, ou la plus grande entrée de celle-ci est bornée par une constante). Nous présentons de nouvelles bornes inférieures sur le nombre de matrices 1C ainsi que de nouvelles heuristiques.<p><p>Finalement, nous terminons par étudier le cas où l'ensemble des matrices 1C ne nous permet pas de décomposer exactement la matrice donnée. Le but est alors de touver une matrice décomposable qui soit aussi proche que possible de la matrice donnée. Après avoir examiné certains cas polynomiaux nous prouvons que le cas général est difficile à approximer avec une erreur additive de O(mn) où m et n représentent les dimensions de la matrice donnée. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished

京都大学 / 新制・課程博士 / 博士(医学) / 甲第23081号 / 医博第4708号 / 新制||医||1049(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 中本 裕士, 教授 小川 誠司, 教授 武田 俊一 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

京都大学 / 新制・課程博士 / 博士(人間健康科学) / 甲第23121号 / 人健博第83号 / 新制||人健||6(附属図書館) / 京都大学大学院医学研究科人間健康科学系専攻 / (主査)教授 椎名 毅, 教授 藤井 康友, 教授 平井 豊博 / 学位規則第4条第1項該当 / Doctor of Human Health Sciences / Kyoto University / DFAM

<p>Radiation-induced lung injury (RILI) is a common condition in the setting of lung and breast cancer. Often, patients who suffer from RILI experience pneumonitis and pulmonary fibrosis months after treatment. These pathologies have commonly been modeled using mice and observing their deterioration until mortality and quantifying pathology on histological sections.<b> </b></p> <p>With this study, we used a longitudinal microCT and a 7T MRI to characterize male C57Bl/6 mice irradiated with a single dose of 20 Gy to the whole thoracic area delivered by an X-Rad cabinent irradiator. CT was performed with a respiratory gating sequence at 2 week timepoints to construct an RIPF model. The fraction of RIPF to total lung volume was calculated at each time point from images, and the data was anaylzed using one-way ANOVA Welch and Dunnett’s T3 multiple comparisons tests. Tidal lung volumes were also calculated and anlyazed in a simlar manner. Mice were then imaged using MRI and CT at 0, 5, and 8 week timepoints to compare results. These results were analyzed for comparison (ANOVA and Dunnett’s T3) and correlation (Pearson’s r) with each other. Histology was later performed using H&E and Trichrome stains to provide ex-vivo verification of pathology.<b> </b>At the 10-12 week time point ( ) significant RIPF formed. Weeks proceeding showed increased significance until the 22+ week timepoint, which showed less statistical significance ( ) due to increased variance at this timepoint. Dunnett’s T3 test showed no significant differences between tidal lung volumes over time. Tests also showed no significant differences between CT and MRI results with a correlation coefficient of .<b> </b>Early in the study, problems arose when pre-marture mortality was occurring to a significant portion of our subjects. Analysis later showed issues during irradiation that resulted in significant dose being absorbed by the stomach. Adjusting our shiedling lead to increased early survival of our subjects enabling us to contine our study. Significant RIPF development was not significant until 10-12 weeks post-irradiation, then RIPF became more severe at proceeding timepoints. Tidal lung volume showed no significant deviation over the development of RIPF. This result is most likely affected by the variation of results at later timepoints, since several mice with severe RIPF were significantly hindered in their ability to breathe during the study. MRI results showed close correlation with CT results and prodcued similar values at early timepoints. However, noticeable differences were seen at later timepoints when significant RIPF developed ( ). <b></b></p>

Owing to the steep dose fall-off curves of high precision and accuracy radiation therapy (RT) modalities such as stereotactic body RT (SBRT), treatment plans with extraordinarily small margins to organs at risk (OARs), such as the spinal cord, has been made possible. With this development, patient movements during treatment, i.e. intrafraction motion (IFM), must be monitored more closely. This master thesis was aimed at developing an A-mode ultrasound prototype to detect the motions of the cervical spine as part of an IFM management (IFMM) system. Current IFMM systems have several drawbacks, including invasiveness and indirect measurements. The existing prototype was tested in order to identify areas of improvement. The prototype developed was equipped with a preconditioning circuit that retains the frequency information of the signal. Furthermore, software was developed based on wavelet filtering and enveloping using the Hilbert transform. Multiple logic algorithms were added in order to handle lost signals, competing echoes, echoes from soft tissues etc. The newly-developed prototype was found to have higher accuracy and precision than the pre-existing prototype. It was also more robust when measuring distance to the spine. A difficulty in segmenting the echo for bone arises for low quality signals. Therefore a compromise exists between setup time, including probe adjustment, and signal quality. Future work includes the manufacturing of a new neck rest to enable robust probe adjustment and fixation.

Radiation therapy is afflicted by a multitude of geometric uncertainties, which must be compensated to ensure treatment success. Such mitigation is currently achieved by enlarging the apparent target volume by various safety margins. This thesis investigated uncertainty sources relating to target position and extent in oesophageal carcinoma, both static and dynamic, and evaluated their impact in a combined model. The first were errors inherent to delineation of the gross tumour volume (GTV), where computed tomography (CT) imaging, the overall modality of choice for target volume delineation (TVD), has a tendency to overestimate target extent. Two rival modalities, [18F]-fluorodeoxyglucose positron emission tomography (FDG-PET) and endoscopic ultrasound (EUS), are generally expected to yield more accurate assessments. EUS has previously suffered from a difficulty in transferring its findings to the spatial domain in which TVD is undertaken. This limitation was overcome here through the use of endoscopically implanted fiducial markers visible on the planning CT. This has enabled their inclusion in TVD and allowed a direct comparison of FDG-PET and EUS based target extents, which were found to agree quite well on average, but showed occasional discrepancies on the order of a few cm. Recently published reports on inter-observer variability (IOV) in TVD of oesophageal carcinoma were summarised with a particular focus on its reduction afforded by the use of fiducial markers. The influence of IOV was investigated more widely in other tumour entities, where it was shown to increase during the course of treatment, mostly due to differing adaptation practices. Microscopic disease extension (MDE), undetectable prior to treatment with current imaging techniques, constituted the second uncertainty source. Reports on histopathological measurements of MDE incidence and its distance from the main tumour were reviewed and spatial measurements extracted to derive a combined estimate of the distribution of extension distances. The overall incidence was estimated as 14.6%, with individual studies reporting widely differing values. Conventional margin widths to compensate for MDE were extracted from the pooled distribution and found to largely agree with the common clinical choice of 3–5 cm, but associated with broad confidence intervals. The addition of such margins to the GTV defines the clinical target volume (CTV). Most studies acknowledged tissue deformations as a major problem, but not all implemented means to prevent or correct it. Preliminary measurements on oesophageal resection specimens were presented, wherein fiducial markers were used to measure tissue deformations, and might ultimately be used to correct spatial measurements of MDE. Fiducial markers also facilitated the study of inter-fractional target mobility in a cohort (n=23) receiving daily orthogonal X-ray imaging for target positioning verification. Markers were found capable of detecting target misalignments, which were a common occurrence with 54% and 15% of analysed markers and treatment fractions showing displacements from their planned position in excess of 5 and 10mm, respectively. Mobility amplitudes were highest in the longitudinal direction and a dependence on tumour location was hinted at, with motion more restricted for proximally located lesions. Measures of systematic and random mobility components were extracted to derive safety margins, which are added to the CTV to form the planning target volume (PTV). A radiobiological model of tumour control probability (TCP) was then evaluated under different uncertainty scenarios. It simplified the tumour system to its longitudinal dimension, which is most affected by the aforementioned phenomena, and simulated positional uncertainties, as well as MDE. The differential impact of systematic and random mobility components on TCP was demonstrated and margin widths sufficient to limit TCP reduction to 1% could best be described by a quadratic combination of their magnitudes. This composition was still applicable when MDE was introduced and mitigated by a conventional margin, but the relative impact of both components shifted. The addition of a PTV margin to the CTV afforded the MDE-positive subpopulation similar protection against positional uncertainties as the same margin achieved without consideration of MDE. The MDE-negative subpopulation attained a much improved tolerance to positional uncertainties through the CTV margin, which also propagated to the overall population. An alternative mitigation of MDE was attempted by optimising the applied dose distribution to an assumed tumour cell density distribution motivated by the literature, which decreases towards the target edge. The optimisation maximised TCP while preserving the integral dose delivered with a conventional margin, under the assumption that this translates into a similar likelihood of normal tissue toxicity. Reduced doses could be delivered to lower cell density regions without sacrificing overall TCP, but this reduction was modest despite vastly diminished cell densities. When this spared dose was redistributed so as to enlarge the treated area, negligible TCP change was observed, but redistribution to the central target did result in appreciably improved TCP in both subpopulations. These effects persisted when positional uncertainties were added and when MDE incidence was increased to the most extreme value reported in the literature.

<p>Proton beam therapy has shown great promise for cancer treatment due to its high precision in irradiating tumor volumes. However, due to the massive size and expense of the cyclotrons/synchrotrons needed to accelerate the protons, the widespread use of proton therapy is limited. Laser plasma accelerated (LPA) proton beams may be a potential alternative to conventional proton beams: by shooting an ultraintense, ultrashort pulsed laser at a thin target, a plasma sheath electric field may be formed with the capability of accelerating protons to potentially therapeutic energies in very short distances. In addition to accessibility, there is significant uncertainty in proton range in heterogeneous tissues. Thermoacoustic computed tomographic (TACT) imaging has the potential to provide <em>in vivo</em> dose imaging and range verification to address these uncertainties. TACT measures thermoacoustic waves generated from the absorbed dose and implements a 3D filtered backprojection to reconstruct volumetric images of the dose. The purpose of this thesis is to determine the feasibility of integrating LPA proton beams with thermoacoustic imaging into a novel image-guided small animal therapy platform as an early step towards clinical  translation to address the issues of accessibility and dosimetric spatial uncertainty. A Monte Carlo (MC) method is used to simulate an LPA proton beam with characteristics based on literature, thermoacoustic waves are simulated on a voxel-wise basis of the MC dose, and 3D filtered backprojection is used to reconstruct a volumetric image of the dose. In Specific Aim 1, the dependence of image accuracy on transducer array angular coverage is investigated; in Specific Aim 2, an iterative reconstruction algorithm is implemented to improve image accuracy through increased sampling of projection space when transducer array angular coverage is insufficient; and in Specific Aim 3, the detector sensitivity to dose is determined for several therapeutic endpoints. The work presented in this thesis not only demonstrates the feasibility of integrating LPA and thermoacoustic technologies but necessary design changes to realize a functional small animal platform.</p>

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第19553号 / 医博第4060号 / 新制||医||1012(附属図書館) / 32589 / 京都大学大学院医学研究科医学専攻 / (主査)教授 増永 慎一郎, 教授 富樫 かおり, 教授 武田 俊一 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

京都大学 / 0048 / 新制・課程博士 / 博士(医学) / 甲第20248号 / 医博第4207号 / 新制||医||1020(附属図書館) / 京都大学大学院医学研究科医学専攻 / (主査)教授 鈴木 実, 教授 別所 和久, 教授 大森 孝一 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

This Master thesis focuses on generative models for medical patient data for radiation therapy. The objective with the project is to implement and investigate the characteristics of a Variational Autoencoder applied to this diverse and versatile data. The questions this thesis aims to answer are: (i) whether the VAE can capture salient features of medical image data, and (ii) if these features can be used to compare similarity between patients. Furthermore, (iii) if the VAE network can successfully reconstruct its input and lastly (iv) if the VAE can generate artificial data having a reasonable anatomical appearance. The experiments carried out conveyed that the VAE is a promising method for feature extraction, since it appeared to ascertain similarity between patient images. Moreover, the reconstruction of training inputs demonstrated that the method is capable of identifying and preserving anatomical details. Regarding the generative abilities, the artificial samples generally conveyed fairly realistic anatomical structures. Future work could be to investigate the VAEs ability to generalize, with respect to both the amount of data and probabilistic considerations as well as probabilistic assumptions. / Fokuset i denna masteruppsats är generativa modeller för patientdata från strålningsbehandling. Syftet med projektet är att implementera och undersöka egenskaperna som en “Variational Autoencoder” (VAE) har på denna typ av mångsidiga och varierade data. Frågorna som ska besvaras är: (i) kan en VAE fånga särdrag hos medicinsk bild-data, och (ii) kan dessa särdrag användas för att jämföra likhet mellan patienter. Därutöver, (iii) kan VAE-nätverket återskapa sin indata väl och slutligen (iv) kan en VAE skapa artificiell data med ett rimligt anatomiskt utseende. De experiment som utfördes pekade på att en VAE kan vara en lovande metod för att extrahera framtydande drag hos patienter, eftersom metoden verkade utröna likheter mellan olika patienters bilder. Dessutom påvisade återskapningen av träningsdata att metoden är kapabel att identifiera och bevara anatomiska detaljer. Vidare uppvisade generellt den artificiellt genererade datan, en realistisk anatomisk struktur. Framtida arbete kan bestå i att undersöka hur väl en VAE kan generalisera, med avseende på både mängd data som krävs och sannolikhetsteorietiska avgränsningar och antaganden.