Debiasing Counterfactuals in the Presence of Spurious Correlations

Amar Kumar*, Nima Fathi, Raghav Mehta, Brennan Nichyporuk, Jean Pierre R. Falet, Sotirios Tsaftaris, Tal Arbel

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

Deep learning models can perform well in complex medical imaging classification tasks, even when basing their conclusions on spurious correlations (i.e. confounders), should they be prevalent in the training dataset, rather than on the causal image markers of interest. This would thereby limit their ability to generalize across the population. Explainability based on counterfactual image generation can be used to expose the confounders but does not provide a strategy to mitigate the bias. In this work, we introduce the first end-to-end training framework that integrates both (i) popular debiasing classifiers (e.g. distributionally robust optimization (DRO)) to avoid latching onto the spurious correlations and (ii) counterfactual image generation to unveil generalizable imaging markers of relevance to the task. Additionally, we propose a novel metric, Spurious Correlation Latching Score (SCLS), to quantify the extent of the classifier reliance on the spurious correlation as exposed by the counterfactual images. Through comprehensive experiments on two public datasets (with the simulated and real visual artifacts), we demonstrate that the debiasing method: (i) learns generalizable markers across the population, and (ii) successfully ignores spurious correlations and focuses on the underlying disease pathology.

Original languageEnglish
Title of host publicationClinical Image-Based Procedures, Fairness of AI in Medical Imaging, and Ethical and Philosophical Issues in Medical Imaging - 12th International Workshop, CLIP 2023 1st International Workshop, FAIMI 2023 and 2nd International Workshop, EPIMI 2023, Proceedings
EditorsStefan Wesarg, Cristina Oyarzun Laura, Esther Puyol Antón, Andrew P. King, John S.H. Baxter, Marius Erdt, Klaus Drechsler, Moti Freiman, Yufei Chen, Islem Rekik, Roy Eagleson, Aasa Feragen, Veronika Cheplygina, Melani Ganz-Benjaminsen, Enzo Ferrante, Ben Glocker, Daniel Moyer, Eikel Petersen
PublisherSpringer Science and Business Media Deutschland GmbH
Pages276-286
Number of pages11
ISBN (Print)9783031452482
DOIs
Publication statusPublished - 9 Oct 2023
Event12th International Workshop on Clinical Image-Based Procedures, CLIP 2023, 1st MICCAI Workshop on Fairness of AI in Medical Imaging, FAIMI 2023, held in conjunction with MICCAI 2023 and 2nd MICCAI Workshop on the Ethical and Philosophical Issues in Medical Imaging, EPIMI 2023 - Vancouver, Canada
Duration: 12 Oct 202312 Oct 2023

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume14242 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference12th International Workshop on Clinical Image-Based Procedures, CLIP 2023, 1st MICCAI Workshop on Fairness of AI in Medical Imaging, FAIMI 2023, held in conjunction with MICCAI 2023 and 2nd MICCAI Workshop on the Ethical and Philosophical Issues in Medical Imaging, EPIMI 2023
Country/TerritoryCanada
CityVancouver
Period12/10/2312/10/23

Keywords / Materials (for Non-textual outputs)

  • Biomarker
  • Counterfactuals
  • Debiasing
  • Explainablity

Fingerprint

Dive into the research topics of 'Debiasing Counterfactuals in the Presence of Spurious Correlations'. Together they form a unique fingerprint.

Cite this