Statistical Shape Analysis of Neuroanatomical Structures via Level-Set--based Shape Morphing

Abstract

Groupwise statistical analysis of the morphometry of brain structures plays an important role in neuroimaging studies. Nevertheless, most morphometric measurements are often limited to volume and surface area, as further morphological characterization of anatomical structures poses a significant challenge. In this paper, we present a method that allows the detection, localization, and quantification of statistically significant morphological differences in complex brain structures between populations. This is accomplished by a novel level-set framework for shape morphing and a multishape dissimilarity-measure derived by a modified version of the Hausdorff distance. The proposed method does not require explicit one-to-one point correspondences and is fast, robust, and easy to implement regardless of the topological complexity of the anatomical surface under study. The proposed model has been applied to well-defined regions of interest using both synthetic and real data sets. This includes the corpus callosum, striatum, caudate, amygdala-hippocampal complex, and superior temporal gyrus. These structures were selected for their importance with respect to brain regions implicated in a variety of neurological disorders. The synthetic databases allowed quantitative evaluations of the method. Results obtained with real clinical data of Williams syndrome and schizophrenia patients agree with published findings in the psychiatry literature. A preliminary version of this paper appeared as Statistical shape analysis for population studies via level-set based shape morphing in the proceedings Computer Vision-ECCV 2012.interest, the morphometric measures are often limited to volume and surface area, as well as other quantitative measures such as curvature, smoothness, and thickness. Yet, these features provide only a partial description of the anatomy and are often insufficient to define a clear distinction between populations.Deformation-based morphometry (DBM) methods, e.g., tensor-based morphometry (TBM), and voxel-based morphometry (VBM) are image analysis tools for identifying regional structural differences from the gradients (Jacobian matrices) of deformation fields [1,16]. In its classical form, VBM does not refer explicitly to specific anatomical structures but considers images as continuous scalar measurements and tests for local differences at a predefined spatial scale. The art of setting the smoothness kernel, which determines the scale at which spatial anatomical differences can be expressed, is a major limitation in VBM [77,78]. Nevertheless, VBM methods gain their popularity in morphological studies, since they work directly on the grayscale images without having to extract the boundaries of particular structures of interest.The shape of a structure, however, often provides rich anatomical characteristics, which may allow one to distinguish between groups of subjects, and could be related to differences in cognitive functioning, development, or neurological symptoms, for example, in cortical folding studies [7...

Advanced Machine Learning Approaches for Brain Mapping

“…Once the hippocampus is successfully segmented and its fine-scale morphology features are extracted, we need to design a technique that specifically compares the dentated structures underneath the CA1 region. Previously, the analysis of shapes is usually conducted between two groups of shapes, trying to identify the region where the two groups of shapes differ significantly (Gerig et al, 2001;Shen and Makedon, 2006;Styner et al, 2006;Cates et al, 2008;Shen et al, 2009;Shen, 2010;Riklin Raviv et al, 2014;Hong et al, 2015;Gao and Bouix, 2016). However, the scenario is different in this work as we have already identified certain regions on the shape, as well as the possible pattern of variation.…”

Fine scale hippocampus morphology variation cross 552 healthy subjects from age 20 to 80

“…Once the hippocampus is successfully segmented and its fine-scale morphology features are extracted, we need to design a technique that specifically compares the dentated structures underneath the CA1 region. Previously, the analysis of shapes is usually conducted between two groups of shapes, trying to identify the region where the two groups of shapes differ significantly (Gerig et al, 2001 ; Shen and Makedon, 2006 ; Styner et al, 2006 ; Cates et al, 2008 ; Shen et al, 2009 ; Shen, 2010 ; Riklin Raviv et al, 2014 ; Hong et al, 2015 ; Gao and Bouix, 2016 ). However, the scenario is different in this work as we have already identified certain regions on the shape, as well as the possible pattern of variation.…”