Brain Structures Segmentation by using Statistical Models

This paper presents an application of computer vision methods to dental age estimation based on the lower third right molar in panoramic X-ray images. For this purpose, two statistical computer vision models are adjusted and applied: Active Shape Model and Active Appearance Model. Both models use shape and appearance of the object to find the outer contour, with the only difference being in the way appearance is used. Statistical models are used to extract features describing the selected tooth, and neural network is used to provide dental age estimation using the features as input. Our own dataset was created, consisting of panoramic X-ray images with known age. A manual segmentation of the selected tooth has been performed for each image in the training set, and the obtained outer contours were used to train both models. Promising preliminary results are presented.

Pose robust face tracking by combining view-based AAMs and temporal filters

Segmentation of diseased lungs in CT images is a nontrivial problem. As Active Appearance Model (AAM) has been applied effectively in this field, we propose a new approach for the construction of traditional AAM to segment the lung fields more accurately and efficiently: Matrixes based AAM (MatAAM). MatAAM is based on two-dimensional image matrixes rather one-dimensional vectors. Its appearance matrix does not need to be transformed into a vector prior to computing the appearance parameter. Instead, a covariance matrix is constructed directly using the normalized appearance matrixes and its eigenvectors are derived for the appearance parameter. The experiment results were compared to other landmark-based methods: Snake, Active Shape Model (ASM), AAM and several modified versions of them. For segmentation of lungs especially diseased lungs, MatAAM performed a superior result in both precision and efficiency.

The field of medical image analysis has greatly influenced the practice of neuroscience. Many studies aim to find new disease related anatomical characteristics based on the analysis of Magnetic Resonance images (MRI). Image segmentation is usually a fundamental step, in which brain structures of interest are delineated from MRI. Neuroscientists then categorize segmentations of anatomical structures by attributes such as shape, size, or location. Besides medical images, biometric images such as facial images are gaining more and more attention from clinical research because certain biometric features are also related to neurological diseases. Similar to brain morphology analysis, facial features are collected for studies testing the hypothesis that the disease alters the facial shape. Despite the rapid development of image analysis technologies, most clinical practice still relies on laborious manual. The aim of this thesis is to develop a shape modeling framework for brain and facial image analysis, for the purpose of clinical practice. Specifically, it focuses on the following three problems: (1) brain structure segmentation from MRI, where several deformable model based segmentation methods are proposed; (2) quantitative analysis of brain morphology, where methods for directly comparing 3D shapes of brain structures are proposed, and clinical applications are demonstrated; (3) facial morphology analysis, where an automatic facial feature localization method is proposed, and applied to the shape analysis of the eyes.

In this work we automatically segment the left ventricle (LV) in cardiac MR images in the end-diastole (ED) and end-systole (ES) phases using a novel approach that combines statistical and deterministic deformable models. A 3D Active Appearance Model (AAM) is used to segment the ED phase. The AAM texture model is trained on radial samples from gradient magnitude images to make the fitting process faster and more discriminative. A trained ED-to-ES shape correspondence model is used to map a given ED shape to an ES shape. Once the AAM model converges to a shape in ED, the correspondence model is used to get an approximate ES shape. We segment the LV in the ES phase by first fitting a deformable superquadric to the AAM converged shape (in ED) using data range forces and then tracking the LV using image and data range forces (for the ES shape obtained from correspondence model). We test our approach by performing leave-one-out training on a 35 patient datasets. The data comprises 19 normal patients and 16 patients having heart abnormalities (cardiomyopathy and myocardial infarction). The composition makes it a challenging data collection with significant shape variation. The performance of our method is evaluated by measuring the mismatch between automatically segmented and expert delineated contours using the Mean Perpendicular Distance (MPD) and Dice metrics. The average MPD is 2.6mm for ED and 3.7mm for ES (error mostly towards the apex and base). The average Dice is 0.9 for ED and 0.8 for ES. These results show good potential for clinical use.

Feature point detection is an important pre-processing step for quantitative evaluation of facial paralysis. Since the conventional methods such as active shape model (ASM) or active appearance model (AAM) are trained by using normal face and they are not possible to detect the feature points accurately for the face with paralysis. In this paper, we propose an automatic and accurate feature point detection method for quantitative evaluation of facial paralysis using deep convolutional neural networks (DCNN). The proposed method consists of two steps. We first use AAM for initial feature point detection. In the second step, a patch with the detected point at the center is used as an input of DCNN for refinement. Experiments demonstrated that the proposed method can significantly improve the detection accuracy of the conventional AAM.

PurposeRepresentation of anatomy appearance is one of the key problems in medical image analysis. An appearance model represents the anatomies with parametric forms, which are then vectorised for prior learning, segmentation and classification tasks.MethodsWe propose a part-based parametric appearance model we refer to as a deformable appearance pyramid (DAP). The parts are delineated by multi-scale local feature pyramids extracted from an image pyramid. Each anatomy is represented by an appearance pyramid, with the variability within a population approximated by local translations of the multi-scale parts and linear appearance variations in the assembly of the parts. We introduce DAPs built on two types of image pyramids, namely Gaussian and wavelet pyramids, and present two approaches to model the prior and fit the model, one explicitly using a subspace Lucas–Kanade algorithm and the other implicitly using the supervised descent method (SDM).ResultsWe validate the performance of the DAP instances with difference configurations on the problem of lumbar spinal stenosis for localising the landmarks and classifying the pathologies. We also compare them with classic methods such as active shape models, active appearance models and constrained local models. Experimental results show that the DAP built on wavelet pyramids and fitted with SDM gives the best results in both landmark localisation and classification.ConclusionA new appearance model is introduced with several configurations presented and evaluated. The DAPs can be readily applied for other clinical problems for the tasks of prior learning, landmark detection and pathology classification.

Facial feature extraction attracts much research interest due to its importance value in applications including human-computer interaction and visual surveillance. Model-based algorithms such as active shape models and active appearance models are efficient for facial feature extraction in color images. However, the complexity of the images and the local nature of active shape models optimization presents a challenge to the localization problem . We propose a new bottom-up method in model fitting stage of ASM, which decompose the whole face model search into several sub-feature search.lt enables the models to be combined with other feature location methods as to get good initialization and lead to more accurate estimates. We demonstrate the experiment results of successfully applying this schema on facial feature extraction.

The accurate segmentation of subcortical brain structures in magnetic resonance (MR) images is of crucial importance in the interdisciplinary field of medical imaging. Although statistical approaches such as active shape models (ASMs) have proven to be particularly useful in the modeling of multiobject shapes, they are inefficient when facing challenging problems. Based on the wavelet transform, the fully generic multiresolution framework presented in this paper allows us to decompose the interobject relationships into different levels of detail. The aim of this hierarchical decomposition is twofold: to efficiently characterize the relationships between objects and their particular localities. Experiments performed on an eight-object structure defined in axial cross sectional MR brain images show that the new hierarchical segmentation significantly improves the accuracy of the segmentation, and while it exhibits a remarkable robustness with respect to the size of the training set.

Active Appearance Models (AAMs) and Constrained Local Models (CLMs) are classical approaches for model-based image segmentation in medical image analysis. AAMs consist of global statistical models of shape and appearance, are known to be hard to fit, and often suffer from insufficient generalization capabilities in case of limited training data. CLMs model appearance only for local patches and relax or completely remove the global appearance constraint. They are, therefore, much easier to optimize but in certain cases they lack the robustness of AAMs. In this paper, we present a framework for patch-based active appearance modeling, which elegantly combines strengths of AAMs and CLMs. Our models provide global shape and appearance constraints and we make use of recent methodological advances from computer vision for efficient joint optimization of shape and appearance parameters during model fitting. Furthermore, the insufficient generalization abilities of those global models are tackled by incorporating and extending a recent approach for learning representative statistical shape models from small training populations. We evaluate our approach on publicly available chest radiographs and cardiac MRI data. The results show that the proposed framework leads to competitive results in terms of segmentation accuracy for challenging multi-object segmentation problems even when only few training samples are available.

Magnetic Resonance Imaging (MRI) biomarkers of multiple sclerosis (MS), particularly fluid-attenuated inversion-recovery (FLAIR) sequences, have long been investigated. However, advanced analytical methods, capable of fusing results from different MRI modalities, could be more informative to have enabled joint biomarkers. Here we estimate disease duration (DD) in MS subjects (n=47) based on fusion of information from Myelin Water Imaging (MWI), Diffusion Tensor Imaging (DTI), and resting state functional MRI (rsfMRI) modalities, by adapting a joint Multimodal Statistical Analysis Framework. Using this data driven, multimodal, latent variable (LV) approach, common and unique information in each dataset was acquired and their relationship with DD is analyzed through the Least Absolute Shrinkage and Selection Operator (LASSO) regression. The common components between the three modalities, but not the unique components of each modality, accurately predicted DD. To further investigate the regions importance for estimating DD, we separated the data into two groups: “early” and “late” DD depending upon their relationship to the median duration of illness (120 months). In early disease, DTI information in the Right Tapetum and the Fornix jointly with MWI information in the Left Superior Cerebellar Penducle, Right Cerebral Penducle, and Cingulum were most informative. In contrast, rsfMRI demonstrated altered connectivity throughout disease duration. Our results demonstrate the power of multimodal imaging markers in MS.

3D visualisation is being used increasingly to improve the accuracy of biopsies and seed placement for prostate cancer brachytherapy (the implantation of small radioactive ‘seeds’ for radiotherapy). The construction of a 3D model requires the segmentation of 2D Magnetic Resonance (MR) images, which remains a challenging problem. The current practice of manual segmentation requires a very high level of expertise and is time-consuming. In this paper, we propose a new method for creating a 3D model of the prostate from MR images obtained pre-operatively. To do this, an Active Appearance Model (AAM) is created from the image data by grouping the MR images based on the region of the prostate each is taken from. A fast optimisation algorithm is then used to segment a new prostate using the AAM. Our results show that this method is fast and accurate compared with manual segmentation enabling this method to be used intra-operatively.

Active Appearance Models (AAMs) are generative parametric models that have been successfully used in the past to model faces. Anecdotal evidence, however, suggests that the performance of an AAM built to model the variation in appearance of a single person across pose, illumination, and expression (a Person Specific AAM) is substantially better than the performance of an AAM built to model the variation in appearance of many faces, including unseen subjects not in the training set (a Generic AAM). In this paper, we present an empirical evaluation that shows that Person Specific AAMs are, as expected, both easier to build and more robust to fit than Generic AAMs. Moreover, we show that: (1) building a generic shape model is far easier than building a generic appearance model, and (2) the shape component is the main cause of the reduced fitting robustness of Generic AAMs. We then proceed to describe two refinements to Generic AAMs to improve their performance: (1) a refitting procedure to improve the quality of the ground-truth data used to build the AAM and (2) a new fitting algorithm. For both refinements we demonstrate dramatically improved fitting performance. Finally, we evaluate the effect of these improvements on a combined model construction and fitting task.

Magnetic resonance (MR) imaging techniques have revolutionalized in vivo investigation of the human brain through anatomical MR imaging (MRI) and functional MR imaging (fMRI). MRI is increasingly being used for quantification of brain tissues and structures while fMRI experiments are used to find brain regions and neural systems involved in a functional task. Due to the close relationship between structure and function of the brain, this thesis explores the variability of brain structures, activations, and connectivity across individual subjects in order to accurately analyze MRI/fMRI data. Though numerous methods have been proposed for accurate identification of brain structures, none of the techniques have adequately addressed the variability of brain structures among individuals. We develop a novel fully automated approach based on fuzzy templates for segmentation of sub-cortical structures, taking into consideration their variability among individuals. Prior knowledge of structure intensity, spatial location, and relative spatial relationships are extracted automatically as fuzzy templates from a training dataset. The structures in a given brain image are then segmented using the fuzzy templates. The promise of the method is demonstrated in the segmentation of five sub-cortical brain structures. The fact that the hemodynamic response function (HRF) varies across different regions and subjects has been seldomly addressed by earlier methods of detection of ATTENTION: The Singapore

The use of appearance and shape priors in image segmentation is known to improve accuracy; however, existing techniques have several drawbacks. For instance, most active shape and appearance models require landmark points and assume unimodal shape and appearance distributions, and the level set representation does not support construction of local priors. In this paper, we present novel appearance and shape models for image segmentation based on a differentiable implicit parametric shape representation called a disjunctive normal shape model (DNSM). The DNSM is formed by the disjunction of polytopes, which themselves are formed by the conjunctions of half-spaces. The DNSM's parametric nature allows the use of powerful local prior statistics, and its implicit nature removes the need to use landmarks and easily handles topological changes. In a Bayesian inference framework, we model arbitrary shape and appearance distributions using nonparametric density estimations, at any local scale. The proposed local shape prior results in accurate segmentation even when very few training shapes are available, because the method generates a rich set of shape variations by locally combining training samples. We demonstrate the performance of the framework by applying it to both 2-D and 3-D data sets with emphasis on biomedical image segmentation applications.