Exploring Instance-Level Uncertainty for Medical Detection

TL;DR

This paper introduces a method to estimate instance-level uncertainty in bounding-box detection for medical imaging, improving detection performance and decision-making in lung nodule detection tasks.

Contribution

It proposes augmenting a 2.5D detection CNN with bounding-box-level uncertainty estimates, specifically predictive variance and Monte Carlo sample variance, which was previously underexplored in detection tasks.

Findings

Detection accuracy improved from 84.57% to 88.86% using combined uncertainty measures.

Uncertainty estimates enable better operating points than probability thresholds.

Performance further increased to 89.52% with uncertainty-based decision thresholds.

Abstract

The ability of deep learning to predict with uncertainty is recognized as key for its adoption in clinical routines. Moreover, performance gain has been enabled by modelling uncertainty according to empirical evidence. While previous work has widely discussed the uncertainty estimation in segmentation and classification tasks, its application on bounding-box-based detection has been limited, mainly due to the challenge of bounding box aligning. In this work, we explore to augment a 2.5D detection CNN with two different bounding-box-level (or instance-level) uncertainty estimates, i.e., predictive variance and Monte Carlo (MC) sample variance. Experiments are conducted for lung nodule detection on LUNA16 dataset, a task where significant semantic ambiguities can exist between nodules and non-nodules. Results show that our method improves the evaluating score from 84.57% to 88.86% by utilizing a combination of both types of variances. Moreover, we show the generated uncertainty enables superior operating points compared to using the probability threshold only, and can further boost the performance to 89.52%. Example nodule detections are visualized to further illustrate the advantages of our method.