Style Curriculum Learning for Robust Medical Image Segmentation

TL;DR

This paper introduces a style curriculum learning framework that enhances the robustness of medical image segmentation models against distribution shifts, especially in multi-centre studies with varying scanner protocols.

Contribution

It proposes a novel style curriculum with automated gradient control and local gradient sign strategies to improve model robustness without target data.

Findings

Significant improvement in segmentation accuracy on the M&Ms dataset.

Enhanced generalization to unknown data distributions.

Robustness against multi-vendor scanner variations.

Abstract

The performance of deep segmentation models often degrades due to distribution shifts in image intensities between the training and test data sets. This is particularly pronounced in multi-centre studies involving data acquired using multi-vendor scanners, with variations in acquisition protocols. It is challenging to address this degradation because the shift is often not known \textit{a priori} and hence difficult to model. We propose a novel framework to ensure robust segmentation in the presence of such distribution shifts. Our contribution is three-fold. First, inspired by the spirit of curriculum learning, we design a novel style curriculum to train the segmentation models using an easy-to-hard mode. A style transfer model with style fusion is employed to generate the curriculum samples. Gradually focusing on complex and adversarial style samples can significantly boost the robustness of the models. Second, instead of subjectively defining the curriculum complexity, we adopt an automated gradient manipulation method to control the hard and adversarial sample generation process. Third, we propose the Local Gradient Sign strategy to aggregate the gradient locally and stabilise training during gradient manipulation. The proposed framework can generalise to unknown distribution without using any target data. Extensive experiments on the public M\&Ms Challenge dataset demonstrate that our proposed framework can generalise deep models well to unknown distributions and achieve significant improvements in segmentation accuracy.