Previous Weston Collaboration

In the previous Weston PVS project, we made significant improvements to our computational PVS pipeline and developed a Docker container to enhance reproducibility. Additionally, we implemented a parallel version of the pipeline to efficiently process large datasets. These enhancements have allowed us to apply our method in large-scale studies, processing some 60,340 scans across 13 different datasets—each with varying protocols, scanners, and locations.

Our method measures PVS count, PVS volume, and PVS morphology, which includes individual length, diameter, and volume. PVS volume is normalised with the brain region volume. The results from analysing both individual and combined datasets indicate that a larger PVS volume is associated with poorer cognitive function after adjusting for key covariates in cross-sectional analyses.

Longitudinally, larger normalised PVS volumes tend to predict a higher likelihood of any cognitive impairment. This association mirrors the cross-sectional findings but can be attenuated across a range of studies, possibly due to variations in participant populations, other long-term changes, or differences in follow-up durations. The relationship between PVS morphology and cognition is less clear, which could be attributed to the measurement methods used.