Statistical analysis of brain tissue images in the wavelet domain: wavelet-based morphometry

Area of Science:

• Neuroimaging
• Statistical analysis
• Biomedical engineering

Background:

• Wavelet transformation is a powerful tool for signal representation in neuroimaging.
• Previous studies show wavelet methods enhance sensitivity and specificity in functional MRI and PET analysis.
• Standard spatial domain analyses may have limitations in detecting subtle structural differences.

Purpose of the Study:

• To propose and evaluate a wavelet-based morphometry (WBM) method for statistical analysis of structural MRI data.
• To estimate inter-group differences in gray matter using a voxel-based morphometry (VBM) framework.
• To compare the performance of WBM against standard VBM techniques.

Main Methods:

• Implemented WBM within a VBM-style analysis framework.
• Compared gray-matter images of healthy subjects with artificially induced cortical thinning against unaltered controls.
• Evaluated WBM against standard VBM (SPM, FSL) using different spatial normalization (SyN, FNIRT) and smoothing techniques.

Main Results:

• WBM demonstrated comparable or superior performance to standard VBM in detecting structural differences.
• The study investigated the impact of smoothing, filter types (Battle-Lemarié), and resolution levels on WBM results.
• Results indicate WBM's efficacy in identifying morphometric variations between subject groups.

Conclusions:

• Wavelet-based morphometry (WBM) presents a promising alternative methodology for structural MRI analysis.
• WBM offers enhanced capabilities for assessing inter-group structural differences in the brain.
• The findings support WBM as a valuable tool for population-based neuroimaging studies.