Weakly Supervised Medical Image Segmentation With Soft Labels and Noise Robust Loss

TL;DR

This paper introduces a noise-robust loss function and probabilistic labeling approach for medical image segmentation, effectively handling noisy annotations and improving segmentation accuracy in MRI knee scans.

Contribution

It proposes a novel normalized active-passive loss function combined with probabilistic labels derived from multi-rater annotations and anatomical knowledge, enhancing segmentation robustness.

Findings

Improved Dice score by 8% over binary cross-entropy

Enhanced precision by 14% and recall by 22%

Mitigated effects of noisy labels in MRI segmentation

Abstract

Recent advances in deep learning algorithms have led to significant benefits for solving many medical image analysis problems. Training deep learning models commonly requires large datasets with expert-labeled annotations. However, acquiring expert-labeled annotation is not only expensive but also is subjective, error-prone, and inter-/intra- observer variability introduces noise to labels. This is particularly a problem when using deep learning models for segmenting medical images due to the ambiguous anatomical boundaries. Image-based medical diagnosis tools using deep learning models trained with incorrect segmentation labels can lead to false diagnoses and treatment suggestions. Multi-rater annotations might be better suited to train deep learning models with small training sets compared to single-rater annotations. The aim of this paper was to develop and evaluate a method to generate probabilistic labels based on multi-rater annotations and anatomical knowledge of the lesion features in MRI and a method to train segmentation models using probabilistic labels using normalized active-passive loss as a "noise-tolerant loss" function. The model was evaluated by comparing it to binary ground truth for 17 knees MRI scans for clinical segmentation and detection of bone marrow lesions (BML). The proposed method successfully improved precision 14, recall 22, and Dice score 8 percent compared to a binary cross-entropy loss function. Overall, the results of this work suggest that the proposed normalized active-passive loss using soft labels successfully mitigated the effects of noisy labels.