CIT-LR, a diffusion-based statistical test, checks whether medical imaging classifiers change their predictions when a patient's reported race is altered in generated counterfactual images.
Same setting is applied to Chexpert and MIMIC-CXR dataset
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
AI Alignment in Medical Imaging: Unveiling Hidden Biases Through Counterfactual Analysis
CIT-LR, a diffusion-based statistical test, checks whether medical imaging classifiers change their predictions when a patient's reported race is altered in generated counterfactual images.