DOMINO: Domain-aware Loss for Deep Learning Calibration

TL;DR

This paper introduces DOMINO, a domain-aware loss function that improves calibration of deep learning models in medical imaging by applying class-wise penalties based on class similarities, reducing risky errors.

Contribution

The paper presents a novel domain-aware loss function that enhances model calibration and safety in high-risk applications, especially in medical imaging.

Findings

Improved calibration of deep learning models in medical imaging.

Reduced risky errors in model predictions.

Code availability for reproducibility.

Abstract

Deep learning has achieved the state-of-the-art performance across medical imaging tasks; however, model calibration is often not considered. Uncalibrated models are potentially dangerous in high-risk applications since the user does not know when they will fail. Therefore, this paper proposes a novel domain-aware loss function to calibrate deep learning models. The proposed loss function applies a class-wise penalty based on the similarity between classes within a given target domain. Thus, the approach improves the calibration while also ensuring that the model makes less risky errors even when incorrect. The code for this software is available at https://github.com/lab-smile/DOMINO.