Related Experiment Video
Updated: Oct 18, 2025
06:48
Automated Segmentation of Cortical Grey Matter from T1-Weighted MRI Images
Published on: January 7, 2019
9.1K
A deep learning toolbox for automatic segmentation of subcortical limbic structures from MRI images
Douglas N Greve1, Benjamin Billot2, Devani Cordero3
1Athinoula A. Martinos Center for Biomedical Imaging, Department of Radiology, Massachusetts General Hospital, Boston, MA, USA; Harvard Medical School, Radiology Department, Boston, MA, USA.
Neuroimage
|September 27, 2021
Summary
A new tool automatically segments subcortical limbic structures from T1-weighted MRI scans, improving accessibility for clinical research. This automated segmentation shows high accuracy and reliability, aiding in the study of neurological conditions.
Area of Science:
- Neuroimaging
- Medical Image Analysis
- Computational Neuroscience
Background:
- Subcortical limbic structures are crucial for emotion and memory.
- Manual segmentation of these structures is time-consuming and requires expertise.
- Limited automated tools exist for segmenting these specific brain regions.
Purpose of the Study:
- To develop and validate an automated tool for segmenting key subcortical limbic structures.
- To provide a publicly available resource for neuroimaging research.
- To assess the tool's performance and reliability in a large dataset.
Main Methods:
- A U-Net deep learning model was trained using 39 manually labeled T1-weighted MRI datasets.
- Data augmentation techniques (spatial, intensity, contrast, noise) were employed.
- The tool was evaluated on a diverse dataset of 698 subjects.
Main Results:
- The tool achieved high accuracy, with excellent test-retest reliability and failure rates below 1%.
- Automated segmentation volumes correlated well with manual segmentations.
- Segmented volumes effectively detected clinical Alzheimer's Disease and age-related effects.
Conclusions:
- The developed automated tool provides accurate and reliable segmentation of subcortical limbic structures.
- This tool addresses a significant unmet need in neuroimaging research.
- Integration with FreeSurfer will enable comprehensive, automated analysis of the limbic system.
