Efficient Model Monitoring for Quality Control in Cardiac Image Segmentation

TL;DR

This paper introduces a novel framework for monitoring cardiac image segmentation models' performance without ground truth, using anomaly detection to ensure high-quality results in clinical settings.

Contribution

It proposes a new anomaly detection-based monitoring framework with global and pixel-wise quality measures for cardiac segmentation models, enabling ground truth-free performance assessment.

Findings

Framework accurately reproduces challenge rankings

It is fast and scalable for clinical use

Effective in flagging suspicious segmentation results

Abstract

Deep learning methods have reached state-of-the-art performance in cardiac image segmentation. Currently, the main bottleneck towards their effective translation into clinics requires assuring continuous high model performance and segmentation results. In this work, we present a novel learning framework to monitor the performance of heart segmentation models in the absence of ground truth. Formulated as an anomaly detection problem, the monitoring framework allows deriving surrogate quality measures for a segmentation and allows flagging suspicious results. We propose two different types of quality measures, a global score and a pixel-wise map. We demonstrate their use by reproducing the final rankings of a cardiac segmentation challenge in the absence of ground truth. Results show that our framework is accurate, fast, and scalable, confirming it is a viable option for quality control monitoring in clinical practice and large population studies.