Assessing age-related gray matter decline with voxel-based morphometry depends significantly on segmentation and...

Area of Science:

• Neuroimaging
• Brain Aging Research
• Computational Anatomy

Background:

• Healthy aging involves progressive gray matter (GM) decline across the brain.
• Manual region of interest (ROI) volumetry was the traditional method for assessing GM degeneration.
• Voxel-Based Morphometry (VBM) offers automated analysis but depends on accurate image segmentation and normalization.

Purpose of the Study:

• To compare volumetric measurements of age-related GM decline using different VBM algorithms.
• To assess the impact of segmentation and normalization methods on VBM results.
• To evaluate differences between VBM and atlas-based ROI approaches.

Main Methods:

• Compared SPM5/SPM8 unified segmentation/normalization, DARTEL (SPM8), and FSL tools (FAST, FNIRT).
• Utilized an atlas-based ROI approach for volumetric analysis.
• Analyzed brain gray matter changes in young (mean age 23) and elderly (mean age 66) subjects.

Main Results:

• Significant differences in volumetric measurements were observed between methods.
• SPM5/SPM8 showed the largest age-related GM differences; DARTEL showed the smallest.
• Method-specific differences were most pronounced in cortical areas near major sulci.

Conclusions:

• Algorithms with limited local deformation flexibility (e.g., SPM5/SPM8 unified segmentation) may overestimate group differences in VBM.
• Flexible warping methods (e.g., DARTEL, FSL) provide more conservative estimates.
• Findings are crucial for comparing VBM results across studies using different methodologies, especially when group anatomy deviates from templates.