An open-source tool for longitudinal whole-brain and white matter lesion segmentation

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Neurology

Background:

• Longitudinal Magnetic Resonance Imaging (MRI) is crucial for tracking brain changes in neurodegenerative diseases.
• Existing whole-brain segmentation methods may lack the temporal consistency needed for subtle change detection.
• White matter lesions require robust analysis in longitudinal studies.

Purpose of the Study:

• To develop and validate a longitudinal whole-brain segmentation method for MRI scans.
• To enhance tracking of subtle morphological changes in neuroanatomical structures and white matter lesions.
• To improve sensitivity to longitudinal disease effects in patient cohorts.

Main Methods:

• Extension of an existing multi-contrast whole-brain segmentation method.
• Incorporation of subject-specific latent variables for temporal consistency.
• Validation on datasets of control subjects and patients with Alzheimer's disease and multiple sclerosis.

Main Results:

• The proposed longitudinal method demonstrates higher test-retest reliability compared to cross-sectional and benchmark methods.
• The method shows increased sensitivity to longitudinal disease effects differentiating patient groups.
• Accurate segmentation of neuroanatomical structures and white matter lesions over time.

Conclusions:

• The developed longitudinal segmentation method offers improved accuracy and reliability for neuroimaging research.
• This technique aids in the early detection and monitoring of neurodegenerative diseases.
• The method is available as part of the open-source FreeSurfer package.