Runtime Freezing: Dynamic Class Loss for Multi-Organ 3D Segmentation

TL;DR

This paper introduces dynamic class loss strategies to improve multi-organ 3D segmentation performance, addressing class imbalance and data scarcity issues in medical imaging.

Contribution

The paper proposes a novel dynamic class loss method that adapts during training to handle class imbalance in multi-organ segmentation tasks.

Findings

Improved segmentation accuracy on a challenging abdominal dataset.

Effective mitigation of class imbalance issues.

Enhanced robustness of segmentation models in medical imaging.

Abstract

Segmentation has become a crucial pre-processing step to many refined downstream tasks, and particularly so in the medical domain. Even with recent improvements in segmentation models, many segmentation tasks remain difficult. When multiple organs are segmented simultaneously, difficulties are due not only to the limited availability of labelled data, but also to class imbalance. In this work we propose dynamic class-based loss strategies to mitigate the effects of highly imbalanced training data. We show how our approach improves segmentation performance on a challenging Multi-Class 3D Abdominal Organ dataset.