Mindboggling morphometry of human brains

Arno Klein, Satrajit Ghosh, Forrest Sheng Bao, Joachim Giard, Yrjö Häme, Eliezer Stavsky, Noah Lee, B Rossa, Martin Reuter, Elias Chaibub Neto, Anisha Keshavan

TL;DR

This paper introduces Mindboggle, an open-source platform for detailed human brain shape analysis using MRI data. It offers a wide array of shape measures beyond traditional volume and thickness, enabling more sensitive detection of morphological differences relevant to mental health disorders.

What Problem Does This Paper Address?

Understanding the human brain's shape and how it varies is crucial for diagnosing and tracking neurological and psychiatric conditions. However, existing methods primarily rely on simple volumetric measures like volume and cortical thickness. These are often insufficient to capture the subtle, complex shape variations that might serve as more sensitive biomarkers. Furthermore, comparing brain shapes across individuals or over time is challenging due to anatomical variability and the need for accurate correspondence. This paper addresses the need for a comprehensive, open-source tool that can extract a richer set of shape features from MRI data to improve our ability to study brain morphology and its relation to health and disease.

Methods

Mindboggle is a modular, Python-based platform built on the Nipype framework. It processes preprocessed T1-weighted MRI data to output volume, surface, and tabular data containing labels, features, and shape information. Key methodological contributions include:

  • Hybrid Segmentation: Combining FreeSurfer and ANTs segmentation outputs to improve accuracy in gray/white matter boundary delineation.
  • Novel Shape Measures: Development of new algorithms for computing surface-based shape measures, including geodesic depth and travel depth, which are more sensitive to subtle morphological changes than traditional methods.
  • Comprehensive Feature Extraction: Implementation of algorithms to extract various shape features at different levels:
  • Volumetric: Volume and a novel thickness measure (thickinthehead) that doesn't rely on surface reconstruction.
  • Surface Vertex-wise: Surface area, mean curvature, geodesic depth, and travel depth.
  • Surface Region-wise: Laplace-Beltrami spectra and Zernike moments.
  • Hierarchical Features: Extraction of folds and sulcal fundi.
  • Open Source and Reproducibility: The software is open-source, with code available on GitHub and extensive documentation, promoting transparency and reproducibility.

Datasets

The primary dataset used for evaluation and demonstration is the Mindboggle-101 dataset. This dataset consists of 101 manually labeled human brain MRIs, described as the largest set of manually labeled, publicly available brain images at the time of publication. The paper also mentions using data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) for Alzheimer's disease-related analyses.

Key Insights and Ideas

  • Subtle Shape Matters: The paper strongly advocates for the importance of "subtle shape measures" beyond volume and thickness, suggesting they can provide more sensitive and specific biomarkers for mental health disorders.
  • Travel Depth's Advantage: Mindboggle's travel depth measure is shown to be more faithful than FreeSurfer's convexity for representing shallow portions of the brain surface, offering a more nuanced depth assessment.
  • Inter-hemispheric Differences are Significant: Shape differences between the left and right hemispheres are substantially larger than scan-rescan differences within the same hemisphere, highlighting the importance of considering hemispheric asymmetry.
  • Variability is Primarily Residual: For most shape measures and brain regions, the largest source of variability is the "residual" component, suggesting that individual differences within a subject's hemispheres are more dominant than differences due to laboratory, subject, or hemisphere.

Key Results

  • Comprehensive Analysis: Mindboggle enables the largest and most detailed shape analysis of human brains conducted to date, utilizing a wide array of shape measures and a substantial population.
  • Consistency of Measures: Mindboggle's thickinthehead cortical thickness measure shows good consistency across scans and generates values closer to published ranges compared to FreeSurfer or ANTs.
  • Correlation of Shape Measures: The study found that while many shape measures are correlated (e.g., travel and geodesic depth), they are not entirely independent, emphasizing the need for careful interpretation when using one as a proxy for another.
  • Hemispheric Asymmetry: Shape differences between hemispheres are significantly larger than scan-rescan differences, with values often exceeding 0.25 and sometimes reaching over 1.00 for certain regions and measures.
  • Alzheimer's Disease Correlation: Changes in specific shape measures (volume, FreeSurfer thickness, mean curvature) over a three-year interval were significantly correlated with changes in cognitive scores in ADNI participants with Alzheimer's disease.

Limitations and Future Work

  • Limitations:
    • The paper acknowledges that some shape measures are not independent and can exaggerate values for certain structures (e.g., geodesic depth for the insula).
  • The evaluation of fundus extraction algorithms did not yield a clear winner, with some algorithms extracting too few features for adequate comparison.
  • Very small median curvature values can inflate fractional difference calculations.
  • Future Work:
    • Further evaluations of Mindboggle's algorithms on new datasets, including those with different imaging parameters, genetic information, and higher field strengths.
  • Expanding feature extraction to include sulcal pits and other brain image features.
  • Developing multivariate analyses and independent corroboration of morphology.
  • Exploring deep learning approaches to automate feature extraction and labeling, potentially reducing reliance on handcrafted features and improving performance on pathological conditions.
  • Investigating interactions among shape measures to understand higher-order morphological relationships.
  • Assessing the impact of scan differences on morphometry-based clinical research.

Summary updated

Abstract

Abstract Mindboggle ( http://mindboggle.info ) is an open source brain morphometry platform that takes in preprocessed T1-weighted MRI data and outputs volume, surface, and tabular data containing label, feature, and shape information for further analysis. In this article, we document the software and demonstrate its use in studies of shape variation in healthy and diseased humans. The number of different shape measures and the size of the populations make this the largest and most detailed shape analysis of human brains every conducted. Brain image morphometry shows great potential for providing much-needed biological markers for diagnosing, tracking, and predicting progression of mental health disorders. Very few software algorithms provide more than measures of volume and cortical thickness, and more subtle shape measures may provide more sensitive and specific biomarkers. Mindboggle computes a variety of (primarily surface-based) shapes: area, volume, thickness, curvature, depth, Laplace-Beltrami spectra, Zernike moments, etc. We evaluate Mindboggle’s algorithms using the largest set of manually labeled, publicly available brain images in the world and compare them against state-of-the-art algorithms where they exist. All data, code, and results of these evaluations are publicly available. Author Summary Brains vary in many ways, including their shape. Analysing differences in shape between brains or changes in brain shape over time has been used to characterize morphology of diseased brains, but these analyses conventionally rely on simple volumetric shape measures. We believe that access to a greater variety of shape measures could provide greater sensitivity and specificity to morphological disturbances, and could aid in diagnosis, tracking, and prediction of the progression of mental health disorders. Mindboggle is open source software that provides neuroscientists (and indeed, anyone interested in computing shapes) tools for computing a variety of shape measures, including area, volume, thickness, curvature, geodesic depth, travel depth, Laplace-Beltrami spectra, and Zernike moments. In addition to algorithmic contributions, we conducted evaluations and applied Mindboggle to conduct the most detailed shape analysis of human brains.
BibTeX
@article{klein2017mindboggling,
  author = {Arno Klein and Satrajit Ghosh and Forrest Sheng Bao and Joachim Giard and Yrjö Häme and Eliezer Stavsky and Noah Lee and B Rossa and Martin Reuter and Elias Chaibub Neto and Anisha Keshavan},
  title = {{Mindboggling morphometry of human brains}},
  journal = {PLOS Computational Biology},
  year = {2017},
  doi = {10.1371/journal.pcbi.1005350},
}
Figure