Usable Region Estimate for Assessing Practical Usability of Medical Image Segmentation Models

TL;DR

This paper introduces the Usable Region Estimate (URE) and Correctness-Confidence Rank Correlation (CCRC) to quantitatively evaluate the practical usability of medical image segmentation models, focusing on confidence correctness correlation and usable prediction regions.

Contribution

The paper proposes novel metrics, URE and CCRC, to assess the practical usability of segmentation models by measuring confidence correctness correlation and usable prediction regions.

Findings

URE effectively quantifies model usability in practice.

Models with larger usable regions are more practical.

Methods validated on six datasets with strong performance.

Abstract

We aim to quantitatively measure the practical usability of medical image segmentation models: to what extent, how often, and on which samples a model's predictions can be used/trusted. We first propose a measure, Correctness-Confidence Rank Correlation (CCRC), to capture how predictions' confidence estimates correlate with their correctness scores in rank. A model with a high value of CCRC means its prediction confidences reliably suggest which samples' predictions are more likely to be correct. Since CCRC does not capture the actual prediction correctness, it alone is insufficient to indicate whether a prediction model is both accurate and reliable to use in practice. Therefore, we further propose another method, Usable Region Estimate (URE), which simultaneously quantifies predictions' correctness and reliability of confidence assessments in one estimate. URE provides concrete information on to what extent a model's predictions are usable. In addition, the sizes of usable regions (UR) can be utilized to compare models: A model with a larger UR can be taken as a more usable and hence better model. Experiments on six datasets validate that the proposed evaluation methods perform well, providing a concrete and concise measure for the practical usability of medical image segmentation models. Code is made available at https://github.com/yizhezhang2000/ure.