Brain shape regression components.

Area of Science:

• Neuroimaging
• Computational anatomy
• Neurodegeneration

Background:

• Neurodegenerative diseases like Alzheimer's disease (AD) cause brain atrophy.
• Understanding the spatial patterns of atrophy is crucial for disease diagnosis and progression monitoring.
• Existing methods for analyzing brain shape and disease burden have limitations in summarizing shape features or discovering associations.

Purpose of the Study:

• To develop a novel method for identifying associations between brain region shape and clinical measures of neurodegenerative disease burden.
• To overcome limitations of prior methods that either oversimplified shape patterns or ignored clinical relevance during feature extraction.

Main Methods:

• A new method was developed to extract linear shape features, termed shape regression components.
• These components are optimized to capture significant population shape variability while being highly correlated with clinical variables.
• The method was applied to hippocampal shape data from 299 Alzheimer's Disease Neuroimaging Initiative (ADNI) participants.

Main Results:

• The novel method identified significant correlations between hippocampal shape changes and markers of AD pathology and cognition.
• These identified associations were stronger and covered a more extensive spatial region compared to those found by competing methods.
• The findings suggest distinct spatial patterns of hippocampal atrophy associated with Alzheimer's disease.

Conclusions:

• The developed method effectively links brain shape properties to neurodegenerative disease burden.
• This approach offers improved sensitivity and spatial coverage for detecting shape-clinical variable associations in conditions like Alzheimer's disease.
• Shape regression components provide a powerful tool for understanding the neuroanatomical basis of neurodegeneration.