A Bayesian Model of Shape and Appearance for Subcortical Brain Segmentation

FMRIB Centre, Department of Clinical Neurology, University of Oxford, Oxford, UK.
NeuroImage (Impact Factor: 6.36). 02/2011; 56(3):907-22. DOI: 10.1016/j.neuroimage.2011.02.046
Source: PubMed


Automatic segmentation of subcortical structures in human brain MR images is an important but difficult task due to poor and variable intensity contrast. Clear, well-defined intensity features are absent in many places along typical structure boundaries and so extra information is required to achieve successful segmentation. A method is proposed here that uses manually labelled image data to provide anatomical training information. It utilises the principles of the Active Shape and Appearance Models but places them within a Bayesian framework, allowing probabilistic relationships between shape and intensity to be fully exploited. The model is trained for 15 different subcortical structures using 336 manually-labelled T1-weighted MR images. Using the Bayesian approach, conditional probabilities can be calculated easily and efficiently, avoiding technical problems of ill-conditioned covariance matrices, even with weak priors, and eliminating the need for fitting extra empirical scaling parameters, as is required in standard Active Appearance Models. Furthermore, differences in boundary vertex locations provide a direct, purely local measure of geometric change in structure between groups that, unlike voxel-based morphometry, is not dependent on tissue classification methods or arbitrary smoothing. In this paper the fully-automated segmentation method is presented and assessed both quantitatively, using Leave-One-Out testing on the 336 training images, and qualitatively, using an independent clinical dataset involving Alzheimer's disease. Median Dice overlaps between 0.7 and 0.9 are obtained with this method, which is comparable or better than other automated methods. An implementation of this method, called FIRST, is currently distributed with the freely-available FSL package.

Download full-text


Available from: Brian Patenaude, Apr 24, 2014
  • Source
    • "This dataset that contains 14 manually segmented subcortical structures to be used as reference does not include multispectral data [14]. Among the most widely used and freely available software we find two automatic methods: FreeSurfer (Martinos Center for Biomedical Imaging, Charlestown, Massachusetts, USA) [15] and FSL-First (Centre for Functional Magnetic Resonance Imaging of the Brain, Oxford, UK) [16] [17]. Both of these software packages that need only T1w images to achieve the segmentation are often used for comparison in the evaluation of new developed methods. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The aim of this paper is investigate the feasibility of automatically training supervised methods, such as k -nearest neighbor ( k NN) and principal component discriminant analysis (PCDA), and to segment the four subcortical brain structures: caudate, thalamus, pallidum, and putamen. The adoption of supervised classification methods so far has been limited by the need to define a representative training dataset, operation that usually requires the intervention of an operator. In this work the selection of the training data was performed on the subject to be segmented in a fully automated manner by registering probabilistic atlases. Evaluation of automatically trained kNN and PCDA classifiers that combine voxel intensities and spatial coordinates was performed on 20 real datasets selected from two publicly available sources of multispectral magnetic resonance studies. The results demonstrate that atlas-guided training is an effective way to automatically define a representative and reliable training dataset, thus giving supervised methods the chance to successfully segment magnetic resonance brain images without the need for user interaction.
    11/2015; 2015:1-9. DOI:10.1155/2015/764383
  • Source
    • "We used FIRST, a model-based segmentation/registration tool for volume comparison of the subcortical structures of males and females (Patenaude et al. 2011). This approach uses deformable surface meshes specific to subcortical structures, namely the amygdala, caudate nucleus, hippocampus, pallidum, putamen and thalamus. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Effects of gender on grey matter (GM) volume differences in subcortical structures of the human brain have consistently been reported. Recent research evidence suggests that both gender and brain size influences volume distribution in subcortical areas independently. The goal of this study was to determine the effects of the interplay between brain size, gender and age contributing to volume differences of subcortical GM in the human brain. High-resolution T1-weighted images were acquired from 53 healthy males and 50 age-matched healthy females. Total GM volume was determined using voxel-based morphometry. We used model-based subcortical segmentation analysis to measure the volume of subcortical nuclei. Main effects of gender, brain volume and aging on subcortical structures were examined using multivariate analysis of variance. No significant difference was found in total brain volume between the two genders after correcting for total intracranial volume. Our analysis revealed significantly larger hippocampus volume for females. Additionally, GM volumes of the caudate nucleus, putamen and thalamus displayed a significant age-related decrease in males as compared to females. In contrast to this only the thalamic volume loss proved significant for females. Strikingly, GM volume decreases faster in males than in females emphasizing the interplay between aging and gender on subcortical structures. These findings might have important implications for the interpretation of the effects of unalterable factors (i.e. gender and age) in cross-sectional structural MRI studies. Furthermore, the volume distribution and changes of subcortical structures have been consistently related to several neuropsychiatric disorders (e.g. Parkinson's disease, attention deficit hyperactivity disorder, etc.). Understanding these changes might yield further insight in the course and prognosis of these disorders.
    Brain Imaging and Behavior 11/2015; DOI:10.1007/s11682-015-9468-3 · 4.60 Impact Factor
  • Source
    • "A mask of the left hippocampus in each participant's high resolution structural space was created using FSL's FIRST and a hippocampal model based on 336 subjects as a prior [Patenaude et al., 2011]. We focused our analysis on the left hippocampus because of the preferential engagement of left-lateralized hippocampal complex areas during verbal memory tasks [Ryan et al., 2008]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: The hippocampal complex is affected early in Alzheimer's disease (AD). Increasingly, altered functional connectivity of the hippocampus is recognized as an important feature of preclinical AD. Carriers of the APOEɛ4 allele are at an increased risk for AD, which could lead to altered hippocampal connectivity even in healthy older adults. To test this hypothesis, we used a paired-associates memory task to examine differences in task-dependent functional connectivity of the anterior and posterior hippocampus in nondemented APOEɛ4 carriers (n = 34, 18F) and noncarriers (n = 46, 31F). We examined anterior and posterior portions of the hippocampus separately to test the theory that APOEɛ4-mediated differences would be more pronounced in the anterior region, which is affected earlier in the AD course. This study is the first to use a psychophysiological interaction approach to query the context-dependent connectivity of subregions of the hippocampus during a memory task in adults at increased genetic risk for AD. During encoding, APOEɛ4 carriers had lower functional connectivity change compared to baseline between the anterior hippocampus and right precuneus, anterior insula and cingulate cortex. During retrieval, bilateral supramarginal gyrus and right precuneus showed lower functional connectivity change with anterior hippocampus in carriers. Also during retrieval, carriers showed lower connectivity change in the posterior hippocampus with auditory cortex. In each case, APOEɛ4 carriers showed strong negative connectivity changes compared to noncarriers where positive connectivity change was measured. These differences may represent prodromal functional changes mediated in part by APOEɛ4 and are consistent with the anterior-to-posterior theory of AD progression in the hippocampus. Hum Brain Mapp, 2015. © 2015 Wiley Periodicals, Inc.
    Human Brain Mapping 10/2015; DOI:10.1002/hbm.23036 · 5.97 Impact Factor
Show more