Multisite harmonization of diffusion tensor image analysis along the perivascular space using the COMBined...

Area of Science:

• Neuroimaging
• Medical Physics
• Biostatistics

Background:

• Diffusion tensor imaging analysis along the perivascular space (DTI-ALPS) is sensitive to variations from different scanners, sites, and protocols.
• These variations can hinder the reliable comparison of DTI-ALPS metrics across multisite studies, particularly in neurodegenerative diseases like Alzheimer's disease (AD).

Purpose of the Study:

• To apply the COMBined Association Test (COMBAT) for harmonizing DTI-ALPS data acquired across multiple sites and scanners.
• To assess the effectiveness of COMBAT in reducing site-specific variations and improving the diagnostic power of the ALPS index in distinguishing Alzheimer's disease (AD) from cognitively normal (CN) individuals.

Main Methods:

• Utilized multisite diffusion magnetic resonance imaging (dMRI) data from 45 AD patients and 82 CN participants from the AD neuroimaging initiative.
• Calculated the ALPS index from dMRI data and applied COMBAT to harmonize scanner, site, and protocol-related variations.
• Compared the ALPS index between AD and CN groups before and after harmonization using Welch's t-test and Cohen's d; assessed correlations with cognitive scores and PET imaging.

Main Results:

• COMBAT successfully harmonized scanner differences, significantly increasing the discriminative power (Cohen's d) of the ALPS index between AD and CN groups.
• The ALPS index showed significant differences between AD and CN post-harmonization (P < 0.05), unlike before harmonization.
• Harmonization led to higher correlations between the ALPS index and cognitive scores, FDG-PET, and AV45-PET SUVRs.

Conclusions:

• COMBAT effectively eliminates between-scanner, site, and protocol variations in DTI-ALPS analysis.
• This harmonization technique facilitates the reliable use of the ALPS index in large-scale, multi-center studies for Alzheimer's disease research.
• The findings suggest COMBAT can enhance the utility of DTI-ALPS as a biomarker in multicenter neuroimaging research.