Dana Tudorascu

This study evaluates harmonization methods to reduce assay and cohort effects in plasma p-tau217 measurements across two cohorts, finding that quantile mapping is most effective.

This paper adapts Superpixel-ComBat to harmonize unpaired multi-site T1W MRI data in HD, improving image quality and enabling unified VBM analyses across scanners.

This study shows that lower periventricular diffusivity is linked to higher amyloid-beta burden and cognitive decline, especially in APOE4 carriers.

MISPEL is a supervised deep learning method for harmonizing multi-scanner neuroimaging data, outperforming existing methods in reducing scanner-related variability.

This paper compares empirical Bayes and fully Bayesian approaches for harmonizing neuroimaging data, finding empirical Bayes more efficient but underestimating uncertainty, while fully Bayesian better preserves biological information.