Leveraging Uncertainty Estimates for Predicting Segmentation Quality
This work addresses the slow clinical adoption of AI systems in medical imaging by providing tools to detect and manage failures, though it is incremental as it applies existing uncertainty techniques to a specific domain.
The paper tackled the problem of silent failures in deep learning-based medical image segmentation by leveraging uncertainty estimates to predict segmentation quality, resulting in a two-stage architecture that produces spatial uncertainty maps and image-level failure predictions.
The use of deep learning for medical imaging has seen tremendous growth in the research community. One reason for the slow uptake of these systems in the clinical setting is that they are complex, opaque and tend to fail silently. Outside of the medical imaging domain, the machine learning community has recently proposed several techniques for quantifying model uncertainty (i.e.~a model knowing when it has failed). This is important in practical settings, as we can refer such cases to manual inspection or correction by humans. In this paper, we aim to bring these recent results on estimating uncertainty to bear on two important outputs in deep learning-based segmentation. The first is producing spatial uncertainty maps, from which a clinician can observe where and why a system thinks it is failing. The second is quantifying an image-level prediction of failure, which is useful for isolating specific cases and removing them from automated pipelines. We also show that reasoning about spatial uncertainty, the first output, is a useful intermediate representation for generating segmentation quality predictions, the second output. We propose a two-stage architecture for producing these measures of uncertainty, which can accommodate any deep learning-based medical segmentation pipeline.