Many research questions in biological research lead to numerous microscope images that need to be evaluated. Here digital image cytometry, i.e., quantitative, automated or semi-automated analysis of the images is an important rapidly growing discipline. This thesis presents contributions to that field. The work has been carried out in close cooperation with biomedical research partners, successfully solving real world problems. The world is 3D and modern imaging methods such as confocal microscopy provide 3D images. Hence, a large part of the work has dealt with the development of new and improved methods for quantitative analysis of 3D images, in particular fluorescently labeled skeletal muscle cells. A geometrical model for robust segmentation of skeletal muscle fibers was developed. Images of the multinucleated muscle cells were pre-processed using a novel spatially modulated transform, producing images with reduced complexity and facilitating easy nuclei segmentation. Fibers from several mammalian species were modeled and features were computed based on cell nuclei positions. Features such as myonuclear domain size and nearest neighbor distance, were shown to correlate with body mass, and femur length. Human muscle fibers from young and old males, and females, were related to fiber type and extracted features, where myonuclear domain size variations were shown to increase with age irrespectively of fiber type and gender. A segmentation method for severely clustered point-like signals was developed and applied to images of fluorescent probes, quantifying the amount and location of mitochondrial DNA within cells. A synthetic cell model was developed, to provide a controllable golden standard for performance evaluation of both expert manual and fully automated segmentations. The proposed method matches the correctness achieved by manual quantification. An interactive segmentation procedure was successfully applied to treated testicle sections of boar, showing how a common industrial plastic softener significantly affects testosterone concentrations.

Geodesic Tractography Segmentation is the two component approach presented in this thesis for the analysis of imagery in oriented domains, with emphasis on the application to diffusion-weighted magnetic resonance imagery (DW-MRI). The computeraided analysis of DW-MRI data presents a new set of problems and opportunities for the application of mathematical and computer vision techniques. The goal is to develop a set of tools that enable clinicians to better understand DW-MRI data and ultimately shed new light on biological processes. This thesis presents a few techniques and tools which may be used to automatically find and segment major neural fiber bundles from DW-MRI data. For each technique, we provide a brief overview of the advantages and limitations of our approach relative to other available approaches. / Acknowledgements page removed per author's request, 01/06/2014.

Techniques described to date for the reproducible and noninvasive quantification of regional cardiac function have been largely relegated to research settings due to time-consuming and cumbersome image acquisition and analysis. In this thesis, feature tracking algorithms are developed for 2-D+Time cardiac magnetic resonance (CMR) and 3-D+Time cardiac computed tomography (CCT) image sequences that are easily acquired clinically, while emphasising reproducibility and automation in their design. First, a commercially-implemented CMR feature tracking algorithm for the analysis of steady state free precession (SSFP) cine series is evaluated in patients with hypertrophic cardiomyopathy (HCM) and arrhythmogenic right ventricular cardiomyopathy (ARVC), which primarily affect the left ventricle (LV) and right ventricle (RV), respectively, and functional impairment compared with control populations is found in both cases. The limitations of this implementation are then used to guide development of an automated algorithm for the same purpose, making use of fully convolutional neural networks (CNN) for segmentation and spline registration across all frames simultaneously for tracking. This study is performed in the subjects with HCM, and functional impairment is again identified in disease subjects. Finally, as myocardial contraction is inherently a 3-D phenomenon, a technique is developed for quantification of regional function from 3-D+Time functional CCT studies using simultaneous registration of automatically generated Loop subdivision surface models for tracking. This study is performed in canine mongrels, and compared with the current state of the art technique for CCT functional analysis. This work demonstrates the feasibility of automated, reproducible cardiac functional analysis from CMR and CCT image sequences. While work remains to be done in extending the principles demonstrated and modular components described to fully automated whole-heart analysis, it is hoped that this thesis will accelerate the clinical adoption of regional functional analysis.

Changes in retinal structure are related to different eye diseases. Various retinal imaging techniques, such as fundus imaging and optical coherence tomography (OCT) imaging modalities, have been developed for non-intrusive ophthalmology diagnoses according to the vasculature changes. However, it is time consuming or even impossible for ophthalmologists to manually label all the retinal structures from fundus images and OCT images. Therefore, computer aided diagnosis system for retinal imaging plays an important role in the assessment of ophthalmologic diseases and cardiovascular disorders. The aim of this PhD thesis is to develop segmentation methods to extract clinically useful information from these retinal images, which are acquired from different imaging modalities. In other words, we built the segmentation methods to extract important structures from both 2D fundus images and 3D OCT images. In the first part of my PhD project, two novel level set based methods were proposed for detecting the blood vessels and optic discs from fundus images. The first one integrates Chan-Vese's energy minimizing active contour method with the edge constraint term and Gaussian Mixture Model based term for blood vessels segmentation, while the second method combines the edge constraint term, the distance regularisation term and the shape-prior term for locating the optic disc. Both methods include the pre-processing stage, used for removing noise and enhancing the contrast between the object and the background. Three automated layer segmentation methods were built for segmenting intra-retinal layers from 3D OCT macular and optic nerve head images in the second part of my PhD project. The first two methods combine different methods according to the data characteristics. First, eight boundaries of the intra-retinal layers were detected from the 3D OCT macular images and the thickness maps of the seven layers were produced. Second, four boundaries of the intra-retinal layers were located from 3D optic nerve head images and the thickness maps of the Retinal Nerve Fiber Layer (RNFL) were plotted. Finally, the choroidal layer segmentation method based on the Level Set framework was designed, which embedded with the distance regularisation term, edge constraint term and Markov Random Field modelled region term. The thickness map of the choroidal layer was calculated and shown.

abstract: Accurate quantitative information of tumor/lesion volume plays a critical role in diagnosis and treatment assessment. The current clinical practice emphasizes on efficiency, but sacrifices accuracy (bias and precision). In the other hand, many computational algorithms focus on improving the accuracy, but are often time consuming and cumbersome to use. Not to mention that most of them lack validation studies on real clinical data. All of these hinder the translation of these advanced methods from benchside to bedside. In this dissertation, I present a user interactive image application to rapidly extract accurate quantitative information of abnormalities (tumor/lesion) from multi-spectral medical images, such as measuring brain tumor volume from MRI. This is enabled by a GPU level set method, an intelligent algorithm to learn image features from user inputs, and a simple and intuitive graphical user interface with 2D/3D visualization. In addition, a comprehensive workflow is presented to validate image quantitative methods for clinical studies. This application has been evaluated and validated in multiple cases, including quantifying healthy brain white matter volume from MRI and brain lesion volume from CT or MRI. The evaluation studies show that this application has been able to achieve comparable results to the state-of-the-art computer algorithms. More importantly, the retrospective validation study on measuring intracerebral hemorrhage volume from CT scans demonstrates that not only the measurement attributes are superior to the current practice method in terms of bias and precision but also it is achieved without a significant delay in acquisition time. In other words, it could be useful to the clinical trials and clinical practice, especially when intervention and prognostication rely upon accurate baseline lesion volume or upon detecting change in serial lesion volumetric measurements. Obviously, this application is useful to biomedical research areas which desire an accurate quantitative information of anatomies from medical images. In addition, the morphological information is retained also. This is useful to researches which require an accurate delineation of anatomic structures, such as surgery simulation and planning. / Dissertation/Thesis / Doctoral Dissertation Biomedical Informatics 2016

La segmentation d’images anatomiques de muscles striés squelettiques acquises par résonance magnétique nucléaire (IRM) présente un grand intérêt pour l’étude des myopathies. Elle est souvent un préalable nécessaire pour l’étude les mécanismes d’une maladie, ou pour le suivi thérapeutique des patients. Cependant, le détourage manuel des muscles est un travail long et fastidieux, au point de freiner les recherches cliniques qui en dépendent. Il est donc nécessaire d’automatiser cette étape. Les méthodes de segmentation automatique se basent en général sur les différences d’aspect visuel des objets à séparer et sur une détection précise des contours ou de points de repère anatomiques pertinents. L’IRM du muscle ne permettant aucune de ces approches, la segmentation automatique représente un défi de taille pour les chercheurs. Dans ce rapport de thèse, nous présentons plusieurs méthodes de segmentation d’images de muscles, toutes en rapport avec l’algorithme dit du marcheur aléatoire (MA). L’algorithme du MA, qui utilise une représentation en graphe de l’image, est connu pour être robuste dans les cas où les contours des objets sont manquants ou incomplets et pour son optimisation numérique rapide et globale. Dans sa version initiale, l’utilisateur doit d’abord segmenter de petites portions de chaque région de l’image, appelées graines, avant de lancer l’algorithme pour compléter la segmentation. Notre première contribution au domaine est un algorithme permettant de générer et d’étiqueter automatiquement toutes les graines nécessaires à la segmentation. Cette approche utilise une formulation en champs aléatoires de Markov, intégrant une connaissance à priori de l’anatomie et une détection préalable des contours entre des paires de graines. Une deuxième contribution vise à incorporer directement la connaissance à priori de la forme des muscles à la méthode du MA. Cette approche conserve l’interprétation probabiliste de l’algorithme original, ce qui permet de générer une segmentation en résolvant numériquement un grand système linéaire creux. Nous proposons comme dernière contribution un cadre d’apprentissage pour l’estimation du jeu de paramètres optimaux régulant l’influence du terme de contraste de l’algorithme du MA ainsi que des différents modèles de connaissance à priori. La principale difficulté est que les données d’apprentissage ne sont pas entièrement supervisées. En effet, l’utilisateur ne peut fournir qu’une segmentation déterministe de l’image, et non une segmentation probabiliste comme en produit l’algorithme du MA. Cela nous amène à faire de la segmentation probabiliste optimale une variable latente, et ainsi à formuler le problème d’estimation sous forme d’une machine à vecteurs de support latents (latent SVM). Toutes les méthodes proposées sont testées et validées sur des volumes de muscles squelettiques acquis par IRM dans un cadre clinique. / Segmentation of magnetic resonance images (MRI) of skeletal striated muscles is of crucial interest when studying myopathies. Diseases understanding, therapeutic followups of patients, etc. rely on discriminating the muscles in MRI anatomical images. However, delineating the muscle contours manually is an extremely long and tedious task, and thus often a bottleneck in clinical research. Typical automatic segmentation methods rely on finding discriminative visual properties between objects of interest, accurate contour detection or clinically interesting anatomical points. Skeletal muscles show none of these features in MRI, making automatic segmentation a challenging problem. In spite of recent advances on segmentation methods, their application in clinical settings is difficult, and most of the times, manual segmentation and correction is still the only option. In this thesis, we propose several approaches for segmenting skeletal muscles automatically in MRI, all related to the popular graph-based Random Walker (RW) segmentation algorithm. The strength of the RW method relies on its robustness in the case of weak contours and its fast and global optimization. Originally, the RW algorithm was developed for interactive segmentation: the user had to pre-segment small regions of the image – called seeds – before running the algorithm which would then complete the segmentation. Our first contribution is a method for automatically generating and labeling all the appropriate seeds, based on a Markov Random Fields formulation integrating prior knowledge of the relative positions, and prior detection of contours between pairs of seeds. A second contribution amounts to incorporating prior knowledge of the shape directly into the RW framework. Such formulation retains the probabilistic interpretation of the RW algorithm and thus allows to compute the segmentation by solving a large but simple sparse linear system, like in the original method. In a third contribution, we propose to develop a learning framework to estimate the optimal set of parameters for balancing the contrast term of the RW algorithm and the different existing prior models. The main challenge we face is that the training samples are not fully supervised. Specifically, they provide a hard segmentation of the medical images, instead of the optimal probabilistic segmentation, which corresponds to the desired output of the RW algorithm. We overcome this challenge by treating the optimal probabilistic segmentation as a latent variable. This allows us to employ the latent Support Vector Machine (latent SVM) formulation for parameter estimation. All proposed methods are tested and validated on real clinical datasets of MRI volumes of lower limbs.

abstract: Recently, a well-designed and well-trained neural network can yield state-of-the-art results across many domains, including data mining, computer vision, and medical image analysis. But progress has been limited for tasks where labels are difficult or impossible to obtain. This reliance on exhaustive labeling is a critical limitation in the rapid deployment of neural networks. Besides, the current research scales poorly to a large number of unseen concepts and is passively spoon-fed with data and supervision. To overcome the above data scarcity and generalization issues, in my dissertation, I first propose two unsupervised conventional machine learning algorithms, hyperbolic stochastic coding, and multi-resemble multi-target low-rank coding, to solve the incomplete data and missing label problem. I further introduce a deep multi-domain adaptation network to leverage the power of deep learning by transferring the rich knowledge from a large-amount labeled source dataset. I also invent a novel time-sequence dynamically hierarchical network that adaptively simplifies the network to cope with the scarce data. To learn a large number of unseen concepts, lifelong machine learning enjoys many advantages, including abstracting knowledge from prior learning and using the experience to help future learning, regardless of how much data is currently available. Incorporating this capability and making it versatile, I propose deep multi-task weight consolidation to accumulate knowledge continuously and significantly reduce data requirements in a variety of domains. Inspired by the recent breakthroughs in automatically learning suitable neural network architectures (AutoML), I develop a nonexpansive AutoML framework to train an online model without the abundance of labeled data. This work automatically expands the network to increase model capability when necessary, then compresses the model to maintain the model efficiency. In my current ongoing work, I propose an alternative method of supervised learning that does not require direct labels. This could utilize various supervision from an image/object as a target value for supervising the target tasks without labels, and it turns out to be surprisingly effective. The proposed method only requires few-shot labeled data to train, and can self-supervised learn the information it needs and generalize to datasets not seen during training. / Dissertation/Thesis / Doctoral Dissertation Computer Science 2020

Statistical modelling has become a ubiquitous tool for analysing of morphological variation of bone structures in medical images. For radiological images, the shape, relative pose between the bone structures and the intensity distribution are key features often modelled separately. A wide range of research has reported methods that incorporate these features as priors for machine learning purposes. Statistical shape, appearance (intensity profile in images) and pose models are popular priors to explain variability across a sample population of rigid structures. However, a principled and robust way to combine shape, pose and intensity features has been elusive for four main reasons: 1) heterogeneity of the data (data with linear and non-linear natural variation across features); 2) sub-optimal representation of three-dimensional Euclidean motion; 3) artificial discretization of the models; and 4) lack of an efficient transfer learning process to project observations into the latent space. This work proposes a novel statistical modelling framework for multiple bone structures. The framework provides a latent space embedding shape, pose and intensity in a continuous domain allowing for new approaches to skeletal joint analysis from medical images. First, a robust registration method for multi-volumetric shapes is described. Both sampling and parametric based registration algorithms are proposed, which allow the establishment of dense correspondence across volumetric shapes (such as tetrahedral meshes) while preserving the spatial relationship between them. Next, the framework for developing statistical shape-kinematics models from in-correspondence multi-volumetric shapes embedding image intensity distribution, is presented. The framework incorporates principal geodesic analysis and a non-linear metric for modelling the spatial orientation of the structures. More importantly, as all the features are in a joint statistical space and in a continuous domain; this permits on-demand marginalisation to a region or feature of interest without training separate models. Thereafter, an automated prediction of the structures in images is facilitated by a model-fitting method leveraging the models as priors in a Markov chain Monte Carlo approach. The framework is validated using controlled experimental data and the results demonstrate superior performance in comparison with state-of-the-art methods. Finally, the application of the framework for analysing computed tomography images is presented. The analyses include estimation of shape, kinematic and intensity profiles of bone structures in the shoulder and hip joints. For both these datasets, the framework is demonstrated for segmentation, registration and reconstruction, including the recovery of patient-specific intensity profile. The presented framework realises a new paradigm in modelling multi-object shape structures, allowing for probabilistic modelling of not only shape, but also relative pose and intensity as well as the correlations that exist between them. Future work will aim to optimise the framework for clinical use in medical image analysis.

abstract: Traumatic injuries are the leading cause of death in children under 18, with head trauma being the leading cause of death in children below 5. A large but unknown number of traumatic injuries are non-accidental, i.e. inflicted. The lack of sensitivity and specificity required to diagnose Abusive Head Trauma (AHT) from radiological studies results in putting the children at risk of re-injury and death. Modern Deep Learning techniques can be utilized to detect Abusive Head Trauma using Computer Tomography (CT) scans. Training models using these techniques are only a part of building AI-driven Computer-Aided Diagnostic systems. There are challenges in deploying the models to make them highly available and scalable. The thesis models the domain of Abusive Head Trauma using Deep Learning techniques and builds an AI-driven System at scale using best Software Engineering Practices. It has been done in collaboration with Phoenix Children Hospital (PCH). The thesis breaks down AHT into sub-domains of Medical Knowledge, Data Collection, Data Pre-processing, Image Generation, Image Classification, Building APIs, Containers and Kubernetes. Data Collection and Pre-processing were done at PCH with the help of trauma researchers and radiologists. Experiments are run using Deep Learning models such as DCGAN (for Image Generation), Pretrained 2D and custom 3D CNN classifiers for the classification tasks. The trained models are exposed as APIs using the Flask web framework, contained using Docker and deployed on a Kubernetes cluster. The results are analyzed based on the accuracy of the models, the feasibility of their implementation as APIs and load testing the Kubernetes cluster. They suggest the need for Data Annotation at the Slice level for CT scans and an increase in the Data Collection process. Load Testing reveals the auto-scalability feature of the cluster to serve a high number of requests. / Dissertation/Thesis / Masters Thesis Software Engineering 2020

Statistical modelling has become a ubiquitous tool for analysing of morphological variation of bone structures in medical images. For radiological images, the shape, relative pose between the bone structures and the intensity distribution are key features often modelled separately. A wide range of research has reported methods that incorporate these features as priors for machine learning purposes. Statistical shape, appearance (intensity profile in images) and pose models are popular priors to explain variability across a sample population of rigid structures. However, a principled and robust way to combine shape, pose and intensity features has been elusive for four main reasons: 1) heterogeneity of the data (data with linear and non-linear natural variation across features); 2) sub-optimal representation of three-dimensional Euclidean motion; 3) artificial discretization of the models; and 4) lack of an efficient transfer learning process to project observations into the latent space. This work proposes a novel statistical modelling framework for multiple bone structures. The framework provides a latent space embedding shape, pose and intensity in a continuous domain allowing for new approaches to skeletal joint analysis from medical images. First, a robust registration method for multi-volumetric shapes is described. Both sampling and parametric based registration algorithms are proposed, which allow the establishment of dense correspondence across volumetric shapes (such as tetrahedral meshes) while preserving the spatial relationship between them. Next, the framework for developing statistical shape-kinematics models from in-correspondence multi-volumetric shapes embedding image intensity distribution, is presented. The framework incorporates principal geodesic analysis and a non-linear metric for modelling the spatial orientation of the structures. More importantly, as all the features are in a joint statistical space and in a continuous domain; this permits on-demand marginalisation to a region or feature of interest without training separate models. Thereafter, an automated prediction of the structures in images is facilitated by a model-fitting method leveraging the models as priors in a Markov chain Monte Carlo approach. The framework is validated using controlled experimental data and the results demonstrate superior performance in comparison with state-of-the-art methods. Finally, the application of the framework for analysing computed tomography images is presented. The analyses include estimation of shape, kinematic and intensity profiles of bone structures in the shoulder and hip joints. For both these datasets, the framework is demonstrated for segmentation, registration and reconstruction, including the recovery of patient-specific intensity profile. The presented framework realises a new paradigm in modelling multi-object shape structures, allowing for probabilistic modelling of not only shape, but also relative pose and intensity as well as the correlations that exist between them. Future work will aim to optimise the framework for clinical use in medical image analysis.