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: 6.13). 02/2011; 56(3):907-22. DOI: 10.1016/j.neuroimage.2011.02.046
Source: PubMed

ABSTRACT 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.

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: This study aimed to investigate whether magnetization transfer imaging (MTI) parameters of cortical gray and white matter and subcortical gray matter structures differ between subjects enriched for human familial longevity and control subjects to provide a thorough description of the brain phenotype of familial longevity. Moreover, we aimed to describe cerebral ageing effects on MTI parameters in an elderly cohort. All subjects were included from the Leiden Longevity Study and underwent 3 Tesla MTI of the brain. In total, 183 offspring of nonagenarian siblings, who are enriched for familial factors of longevity, were contrasted with 163 environmentally and age-matched controls. No differences in cortical and subcortical gray matter and white matter MTI parameters were found between offspring and control subjects using histogram-based and voxel-wise analyses. Cortical gray matter and white matter MTI parameters decreased with increasing chronological age (all p < 0.001). Decrease of white matter magnetization transfer ratio (MTR) was homogeneous throughout the whole mean white matter skeleton except for parts of the callosal splenium and partly the posterior limb of the internal capsule and superior region of the corona radiata (p < 0.05). Mean MTR of subcortical gray matter structures decreased with increasing age (p amygdala, caudate nucleus and putamen < 0.001; p pallidum = 0.001, p thalamus = 0.002). In conclusion, the brain phenotype of human familial longevity is - at a mean age of 66 years - not characterized by preserved macromolecular brain tissue integrity.
    PLoS ONE 01/2015; 10(3):e0120778. DOI:10.1371/journal.pone.0120778 · 3.53 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Early life trauma exposure represents a potent risk factor for the development of mental illnesses such as anxiety, depression, and posttraumatic stress disorder. Moreover, deleterious consequences of trauma are exacerbated in youth living in impoverished, urban environments. A priori probability maps were used to examine resting-state functional connectivity of the amygdala in 21 trauma-exposed, and 21 age- and sex-matched urban children and adolescents (youth) without histories of trauma. Intrinsic functional connectivity analyses focused on amygdala-medial prefrontal circuitry, a key emotion regulatory pathway in the brain. We discovered reduced negative amygdala-subgenual cingulate connectivity in trauma-exposed youth. Differences between groups were also identified in anterior insula and dorsal anterior cingulate to amygdala connectivity. Overall, results suggest a model in which urban-dwelling trauma-exposed youth lack negative prefrontal to amygdala connectivity that may be critical for regulation of emotional responses. Functional changes in amygdala circuitry might reflect the biological embedding of stress reactivity in early life and mediate enhanced vulnerability to stress-related psychopathology. © The Author (2015). Published by Oxford University Press. For Permissions, please email:
    Social Cognitive and Affective Neuroscience 04/2015; DOI:10.1093/scan/nsv030 · 5.88 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: This review aimed to produce hippocampal atrophy rate estimates from healthy ageing studies as well as control samples from observational studies across the adult lifespan which can be used as benchmarks to evaluate abnormal changes in pathological conditions. The review followed PRISMA guidelines. PUBMED (to February 2014) was searched for longitudinal MRI studies reporting hippocampal atrophy or volume change in cognitively healthy individuals. Titles were screened and non-English, duplicate or irrelevant entries excluded. Remaining record abstracts were reviewed to identify studies for full text retrieval. Full text was retrieved and screened against inclusion/exclusion criteria. Bibliographies and previous reviews were examined to identify additional studies. Data were summarised using meta-analysis and age, segmentation technique and study type were tested as potential moderators using meta-regression. It was hypothesised that population studies would produce higher atrophy rates than clinical observational studies. The systematic search identified 4410 entries and 119 studies were retrieved with 58 failing selection or quality criteria, 30 excluded as multiple reports and 3 studies were unsuitable for meta-analysis. The remaining 28 studies were included in the meta-analysis, n = 3422, 44.65% male, 11735 person-years of follow-up, mean age was 24.50 to 83 years. Mean total hippocampal atrophy for the entire sample was 0.85% per year (95% CI 0.63, 1.07). Age based atrophy rates were 0.38% per year (CI 0.14, 0.62) for studies with mean age < 55 years (n=413), 0.98% (CI 0.27, 1.70) for 55 to < 70 years (n=426), and 1.12% (CI 0.86, 1.38) for >= 70 years (n=2583). Meta-regression indicated age was associated with increased atrophy rates of 0.0263% (CI 0.0146, 0.0379) per year and automated segmentation approaches were associated with a reduced atrophy rate of -0.466% (CI -0.841,-0.090). Population studies were not associated with a significant effect on atrophy. Analyses of 11 studies separately measuring left and right hippocampal atrophy (n=1142) provided little evidence of laterality effects. While no study separately reported atrophy by gender, a number tested for gender effects and 2 studies reported higher atrophy in males. Hippocampal atrophy rates increase with age with the largest increases occurring from midlife onwards. Manual segmentation approaches result in higher measured atrophy rates. Copyright © 2015 Elsevier Inc. All rights reserved.
    NeuroImage 03/2015; DOI:10.1016/j.neuroimage.2015.03.035 · 6.13 Impact Factor

Full-text (2 Sources)

Available from
May 29, 2014