Assessing optimisation of 3T MRI protocol across multi-site, multi-vendor settings for studying mild Traumatic Brain Injury: mTBI Predict Consortium study

Large multi-site, multi-vendor MRI studies offer opportunities to advance clinical practice and neuroscience research by increasing sample sizes, especially in clinical populations. However, variability between scanners or sites can affect the reproducibility of quantitative MRI metrics, which is vital for using MRI beyond visual inspection of structural images. This study aimed to assess the reproducibility of MRI metrics for the mild traumatic brain injury-Predict (mTBI-Predict) consortium, spanning Philips and Siemens scanners across three sites. We conducted a travelling head study (n=1), acquiring repeated measurements from a healthy participant across three sites. Evaluated metrics included cortical thickness and subcortical volume from T1-weighted images, fractional anisotropy (FA) and mean diffusivity (MD) from diffusion-weighted imaging (DWI), temporal signal-to-noise ratio (tSNR) from echo planar imaging, and cerebral blood flow (CBF) from arterial spin labelling (ASL). Image quality metrics (point spread function and motion) were also assessed across all DWI, functional magnetic resonance imaging (fMRI), and pseudo-continuous arterial spin labelling (pCASL) data. Standardised pipelines were used for preprocessing and analysis. Within-subject coefficients of variation (wCV%) and linear regressions assessed repeatability and consistency. Structural measures, including cortical thickness, subcortical volume, FA, and MD, demonstrated excellent reproducibility (wCV ≤5%, R² ≥0.85) within and between-sites. Minimal differences in participant motion were observed across sessions. However, as expected, tSNR and CBF showed greater variability than structural measures, due to their functional nature. Despite this, wCV% remained below 10% for all metrics, a threshold indicating very good agreement for biological measures. These findings align with previous work, demonstrating the feasibility of obtaining comparable MRI measurements across sites for studying brain structure and function. This study lays the groundwork for further evaluation of clinically relevant MRI biomarkers for mTBI.