Harmonisation of Enlarged Perivascular Spaces Quantified from Brain Magnetic Resonance Images acquired at 1.5T vs. 3T – Data from a pilot study

Magnetic resonance imaging (MRI)-visible perivascular spaces (PVS) have been found to be associated with disease phenotypes, indicators of sleep quality, and overall health. However, their role is still difficult to ascertain mainly given limitations of sample size and heterogeneity in their acquisition and analysis. We use brain MRI from 100 adult volunteers, acquired at two different magnetic field strengths with different sequence parameters. We visually scored the PVS burden and, independently, automatically segmented the PVS in each scan blind to any patient and imaging acquisition information. We analysed the quantitative a) raw results, b) results after adjusting for the volume of the region of interest, and c) results after applying ComBat harmonisation. This dataset contains the imaging protocols, the raw PVS measurements/visual scores, the measurements after being adjusted by each method, and the Matlab scripts to harmonise the data and statistically analyse the results.