A conformalized learning of a prediction set with applications to medical imaging classification
This addresses the problem of uncertainty quantification for medical imaging classification, enabling safer deployment in clinics, though it is incremental as it builds on conformal prediction methods.
The paper tackles the challenge of quantifying uncertainty in medical imaging classifiers by presenting an algorithm that modifies any classifier to produce prediction sets containing the true label with a user-specified probability, such as 90%, and demonstrates that it outperforms current approaches by achieving smaller average prediction set sizes while maintaining desired coverage.
Medical imaging classifiers can achieve high predictive accuracy, but quantifying their uncertainty remains an unresolved challenge, which prevents their deployment in medical clinics. We present an algorithm that can modify any classifier to produce a prediction set containing the true label with a user-specified probability, such as 90%. We train a network to predict an instance-based version of the Conformal Prediction threshold. The threshold is then conformalized to ensure the required coverage. We applied the proposed algorithm to several standard medical imaging classification datasets. The experimental results demonstrate that our method outperforms current approaches in terms of smaller average size of the prediction set while maintaining the desired coverage.