Multi-site Study of Diffusion Metric Variability: Characterizing the Effects of Site, Vendor, Field Strength, and...

Area of Science:

• Neuroimaging
• Diffusion Tensor Imaging (DTI)
• Medical Imaging

Background:

• Multi-site clinical trials increasingly utilize MRI-based diffusion-weighted imaging (DWI).
• Variability in scanner vendors, protocols, and parameters (e.g., gradient directions, echo time) poses challenges for data harmonization.
• Ensuring reliable and accurate diffusion metrics across diverse sites is critical for multi-center studies.

Purpose of the Study:

• To investigate the impact of site, scanner vendor, field strength, and echo time on key diffusion tensor imaging (DTI) metrics.
• To evaluate the sensitivity of whole-brain histogram distances for characterizing within- and between-site variability in DTI metrics.

Main Methods:

• Analysis of diffusion tensor imaging (DTI) data from multi-site trials.
• Assessment of two primary diffusion metrics: mean diffusivity (MD) and fractional anisotropy (FA).
• Utilized the distance between whole-brain histograms of FA and MD to quantify variability.

Main Results:

• Site, scanner vendor, field strength, and echo time demonstrably influence DTI metrics (MD and FA).
• Region of interest (ROI) metrics and mean histogram values were found to be insufficiently sensitive to site-specific effects.
• Histogram distance analysis proved sensitive to variations introduced by site, vendor, field strength, and echo time.

Conclusions:

• Variability in DTI metrics due to technical factors in multi-center trials can significantly impact study outcomes.
• Whole-brain histogram distance is a sensitive metric for detecting and characterizing these technical variations.
• These findings highlight the need for careful protocol standardization and validation in multi-site DTI studies.