English

Harmonization and the Worst Scanner Syndrome

Machine Learning 2021-04-22 v2 Computer Vision and Pattern Recognition Image and Video Processing Quantitative Methods Machine Learning

Abstract

We show that for a wide class of harmonization/domain-invariance schemes several undesirable properties are unavoidable. If a predictive machine is made invariant to a set of domains, the accuracy of the output predictions (as measured by mutual information) is limited by the domain with the least amount of information to begin with. If a real label value is highly informative about the source domain, it cannot be accurately predicted by an invariant predictor. These results are simple and intuitive, but we believe that it is beneficial to state them for medical imaging harmonization.

Keywords

Cite

@article{arxiv.2101.06255,
  title  = {Harmonization and the Worst Scanner Syndrome},
  author = {Daniel Moyer and Polina Golland},
  journal= {arXiv preprint arXiv:2101.06255},
  year   = {2021}
}

Comments

Med-NeurIPS 2020 Workshop Paper, updated 4/2021

R2 v1 2026-06-23T22:12:51.086Z