Pooling Image Datasets With Multiple Covariate Shift and Imbalance

arXiv:2403.02598v1 Announce Type: new
Abstract: Small sample sizes are common in many disciplines, which necessitates pooling roughly similar datasets across multiple institutions to study weak but relevant associations between images and disease outcomes. Such data often manifest shift/imbalance in covariates (i.e., secondary non-imaging data). Controlling for such nuisance variables is common within standard statistical analysis, but the ideas do not directly apply to overparameterized models. Consequently, recent work has shown how strategies from invariant representation learning provides a meaningful starting …

abstract arxiv cs.cv cs.lg data datasets disease image image datasets images imaging manifest multiple pooling sample shift small study type variables