Exploring semantic consistency in unpaired image translation to generate data for surgical applications

Danush Kumar Venkatesh, Dominik Rivoir, Micha Pfeiffer, Fiona R. Kolbinger, Marius Distler, Jürgen Weitz, Stefanie Speidel

Abstract

In surgical computer vision applications, data privacy and expert annotation challenges impede the acquisition of labeled training data. Unpaired image-to-image translation techniques have been explored to automatically generate annotated datasets by translating synthetic images into a realistic domain. The preservation of structure and semantic consistency, i.e., per-class distribution during translation, poses a significant challenge, particularly in cases of semantic distributional mismatch.
BibTeX
@article{venkatesh2024exploring,
  author = {Danush Kumar Venkatesh and Dominik Rivoir and Micha Pfeiffer and Fiona R. Kolbinger and Marius Distler and Jürgen Weitz and Stefanie Speidel},
  title = {{Exploring semantic consistency in unpaired image translation to generate data for surgical applications}},
  journal = {Int. J. Comput. Assist. Radiol. Surg},
  year = {2024},
  doi = {10.1007/s11548-024-03079-1},
}
Figure