Improving the Reliability of Semantic Segmentation of Medical Images by Uncertainty Modeling with Bayesian Deep Networks and Curriculum Learning
This addresses the critical need for more reliable medical image segmentation, which is essential for accurate diagnosis and treatment planning, though it appears incremental as it builds on existing Bayesian and curriculum learning techniques.
The paper tackles the problem of unreliable semantic segmentation in medical images by proposing a method that combines Bayesian deep networks with curriculum learning to use uncertainty estimates for resampling training data in high-uncertainty areas. The result is a significant increase in model reliability, demonstrated specifically for iPS cell colony segmentation.
In this paper we propose a novel method which leverages the uncertainty measures provided by Bayesian deep networks through curriculum learning so that the uncertainty estimates are fed back to the system to resample the training data more densely in areas where uncertainty is high. We show in the concrete setting of a semantic segmentation task (iPS cell colony segmentation) that the proposed system is able to increase significantly the reliability of the model.