A Kernel-based Approach to Diffusion Tensor and Fiber Clustering in the Human Skeletal Muscle
نویسندگان
چکیده
In this report, we present a kernel-based approach to the clustering of diffusion tensors in images of the human skeletal muscle. Based on the physical intuition of tensors as a means to represent the uncertainty of the position of water protons in the tissues, we propose a Mercer (i.e. positive definite) kernel over the tensor space where both spatial and diffusion information are taken into account. This kernel highlights implicitly the connectivity along fiber tracts. We show that using this kernel in a kernel-PCA setting compounded with a landmark-Isomap embedding and k-means clustering provides a tractable framework for tensor clustering. We extend this kernel to deal with fiber tracts as input using the multiinstance kernel by considering the fiber as set of tensors centered in the sampled points of the tract. The obtained kernel reflects not only interactions between points along fiber tracts, but also the interactions between diffusion tensors. We give an interpretation of the obtained kernel as a comparison of soft fiber representations and show that it amounts to a generalization of the Gaussian kernel Correlation. As in the tensor case, we use the kernel-PCA setting and k-means for grouping of fiber tracts. This unsupervised method is further extended by way of an atlas-based registration of diffusion-free images, followed by a classification of fibers based on non-linear kernel Support Vector Machines (SVMs) and kernel diffusion. The experimental results on a dataset of diffusion tensor images of the calf muscle of 25 patients (of which 5 affected by myopathies, i.e. neuromuscular diseases) show the potential of our method in segmenting the calf in anatomically relevant regions both at the tensor and fiber level. Key-words: DTI, Diffusion tensor, Fiber, Kernels, Clustering, Human skeletal muscle ∗ This work was carried out while the author was at the Signal Processing Laboratory of the Swiss Federal Institute of Technology, Lausanne (EPFL). in ria -0 03 40 61 3, v er si on 1 21 N ov 2 00 8 Une approche basée sur des noyaux pour le groupement de tenseurs de diffusion et de fibres dans le muscle squelettique Résumé : Dans ce rapport, nous présentons une approche basée sur des noyaux pour le groupement de fibres dans le muscle squelettique. En se basant sur l’intuition physique derrière les tenseurs de diffusion comme moyen de représenter l’incertitude sur la position des molécules d’eau dans les tissus, nous proposons un noyau de Mercer (défini positif) sur l’espace des tenseurs qui tient compte aussi bien de l’information spatiale que de l’information de diffusion. Le noyau met l’accent implicitement sur la connectivité le long des fibres. Nous montrons que l’utilisation de ce noyau pour une analyse en composantes principales combinée à l’algorithme Isomap fournit un cadre pratique pour regrouper les tenseurs. Nous étendons ce noyau pour pouvoir segmenter des fibres en considérant une fibre comme étant un ensemble de tenseurs centrés sur les points échantillonés le long de sa trajectoire. Le noyau obtenu reflète aussi bien les intéractions spatiales que les intéractions entre tenseurs de diffusion. Nous donnons également une interprétation de ce noyau comme étant une fao̧n de comparer des représentations probabilistes des fibres. Nous étendons la méthode pour effectuer une segmentation supervisée à l’aide d’un atlas en utilisant des SVMs non linéaires combinés avec une diffusion base sur un champ de Markov. Les résultats expérimentaux effectués sur les données de mollet de 25 patients (dont 5 atteints de myopathies) montrent le potentiel de notre méthode pour segmenter le muscle en régions anatomiquement significatives. Mots-clés : IRM de diffusion, Tenseur de diffusion, Fibre, Noyaux, Groupement, Muscle squelettique in ria -0 03 40 61 3, v er si on 1 21 N ov 2 00 8 A Kernel-based Approach to Diffusion Tensor and Fiber Clustering in the Human Skeletal Muscle3
منابع مشابه
Clustering of the Human Skeletal Muscle Fibers Using Linear Programming and Angular Hilbertian Metrics
In this paper, we present a manifold clustering method fo the classification of fibers obtained from diffusion tensor images (DTI) of the human skeletal muscle. Using a linear programming formulation of prototype-based clustering, we propose a novel fiber classification algorithm over manifolds that circumvents the necessity to embed the data in low dimensional spaces and determines automatical...
متن کاملManifold-driven Grouping of Skeletal Muscle Fibers
In this report, we present a manifold clustering method for the classification of fibers obtained from diffusion tensor images (DTI) of the human skeletal muscle. To this end, we propose the use of angular Hilbertian metrics between multivariate normal distributions to define a family of distances between tensors that we generalize to fibers. The obtained metrics between fiber tracts encompasse...
متن کاملFiber Tractography and Diffusion Tensor Imaging in Children with Agenesis and Dysgenesis of Corpus Callosum: A Clinico-Radiological Correlation
Background Corpus callosum is the largest commissure in human brain. It consists of tightly packed white matter tracts connecting the two cerebral hemispheres. In this study we aimed to evaluate role of fiber tractography (FT), and diffusion tensor imaging (DTI) in ped...
متن کاملDetermination of Fiber Direction in High Angular Resolution Diffusion Images using Spherical Harmonics Functions and Wiener Filter
Diffusion tensor imaging (DTI) MRI is a noninvasive imaging method of the cerebral tissues whose fibers directions are not evaluated correctly in the regions of the crossing fibers. For the same reason the high angular resolution diffusion images (HARDI) are used for estimation of the fiber direction in each voxel. One of the main methods to specify the direction of fibers is usage of the spher...
متن کاملDiffusion-Tensor MRI Based Skeletal Muscle Fiber Tracking.
A skeletal muscle's function is strongly influenced by the internal organization and geometric properties of its fibers, a property known as muscle architecture. Diffusion-tensor magnetic resonance imaging-based fiber tracking provides a powerful tool for non-invasive muscle architecture studies, has three-dimensional sensitivity, and uses a fixed frame of reference. Significant advances have b...
متن کامل