Automatic morphometry in Alzheimer's disease and mild cognitive impairment

Area of Science:

• Neuroimaging
• Medical Image Analysis
• Alzheimer's Disease Research

Background:

• Alzheimer's Disease Neuroimaging Initiative (ADNI) database provides valuable MRI data.
• Accurate brain segmentation is crucial for understanding neurodegenerative diseases.
• Limited publicly available, anatomically segmented brain datasets exist.

Purpose of the Study:

• To create and release a novel repository of anatomically segmented brain images.
• To analyze morphometric differences between healthy controls and Alzheimer's disease patient groups.
• To validate the consistency and utility of the generated brain segmentations.

Main Methods:

• Utilized T1-weighted MRI data from the ADNI database.
• Applied a multi-atlas based MAPER procedure for segmentation of 83 brain regions.
• Performed visual assessment and quantitative analysis of segmentation consistency and group differences.

Main Results:

• Generated segmentations for 816 subjects across normal, mild cognitive impairment, and Alzheimer's disease groups.
• Demonstrated high self-consistency of segmentations across different MRI field strengths (Jaccard coefficient: 0.802±0.0146).
• Identified significant morphometric differences, particularly in the temporal and parietal lobes, and increased asymmetry in posterior cortical regions.

Conclusions:

• The developed repository offers a valuable resource for ADNI data researchers.
• Segmentation-based morphometry effectively differentiates diagnostic groups in Alzheimer's disease.
• The white-matter hypointensities index is a viable tool for quantifying white-matter disease.