Related Experiment Video
Updated: Mar 17, 2026
11:03
High-resolution In Vivo Manual Segmentation Protocol for Human Hippocampal Subfields Using 3T Magnetic Resonance Imaging
Published on: November 10, 2015
10.0K
Bayesian longitudinal segmentation of hippocampal substructures in brain MRI using subject-specific atlases
Juan Eugenio Iglesias1, Koen Van Leemput2, Jean Augustinack3
1Basque Center on Cognition, Brain and Language, Spain; Translational Imaging Group, University College London, United Kingdom.
Neuroimage
|July 19, 2016
Summary
This study introduces a new longitudinal MRI segmentation method for hippocampal subregions. The advanced technique improves accuracy in detecting Alzheimer's disease (AD) and early cognitive impairment progression.
Area of Science:
- Neuroimaging
- Computational Anatomy
- Neurodegenerative Diseases
Background:
- The hippocampal formation's subregions are crucial for memory and are affected in Alzheimer's disease (AD).
- Accurate segmentation of these subregions in MRI scans is vital for understanding disease progression.
Purpose of the Study:
- To develop and evaluate a novel generative model for longitudinal segmentation of hippocampal subregions.
- To improve the accuracy and sensitivity of detecting atrophy in neurodegenerative diseases like AD.
Main Methods:
- A generative model using subject-specific atlases and Bayesian inference for joint longitudinal segmentation.
- Evaluation on over 4700 MRI scans from the ADNI and MIRIAD datasets.
- Comparison against a standard cross-sectional segmentation approach.
Main Results:
- The longitudinal method showed significantly better test-retest reliability (lower volume differences, higher Dice overlap) than the cross-sectional method.
- The algorithm detected subtle atrophy rate differences between Alzheimer's disease (AD) patients and controls in multiple subregions, missed by the cross-sectional approach.
- It also identified differences between controls and early mild cognitive impairment (eMCI) stages, and between eMCI and AD, with higher significance.
Conclusions:
- The proposed longitudinal segmentation algorithm offers superior accuracy and sensitivity for analyzing hippocampal subregion atrophy in neurodegenerative diseases.
- This method enhances the ability to detect early disease changes and track progression more effectively than cross-sectional approaches.
- The algorithm will be integrated into the open-source FreeSurfer neuroimaging package.
