Segmentation of Parotid Gland Tumors Using Multimodal MRI and Contrastive Learning

TL;DR

This paper introduces a Transformer-based contrastive learning approach with transfer learning for improved parotid gland tumor segmentation in MRI images, achieving significant performance gains over supervised methods.

Contribution

It presents a novel contrastive learning framework with transfer learning for medical image segmentation, focusing on the challenging task of parotid gland tumor segmentation.

Findings

Significant improvement in segmentation metrics (DSC, MPA, MIoU, HD) using contrastive learning.

Contrastive learning mainly enhances the encoder part of the segmentation network.

Exploration of contrastive learning for the decoder part and related challenges.

Abstract

Parotid gland tumor is a common type of head and neck tumor. Segmentation of the parotid glands and tumors by MR images is important for the treatment of parotid gland tumors. However, segmentation of the parotid glands is particularly challenging due to their variable shape and low contrast with surrounding structures. Recently deep learning has developed rapidly, which can handle complex problems. However, most of the current deep learning methods for processing medical images are still based on supervised learning. Compared with natural images, medical images are difficult to acquire and costly to label. Contrastive learning, as an unsupervised learning method, can more effectively utilize unlabeled medical images. In this paper, we used a Transformer-based contrastive learning method and innovatively trained the contrastive learning network with transfer learning. Then, the output model was transferred to the downstream parotid segmentation task, which improved the performance of the parotid segmentation model on the test set. The improved DSC was 89.60%, MPA was 99.36%, MIoU was 85.11%, and HD was 2.98. All four metrics showed significant improvement compared to the results of using a supervised learning model as a pre-trained model for the parotid segmentation network. In addition, we found that the improvement of the segmentation network by the contrastive learning model was mainly in the encoder part, so this paper also tried to build a contrastive learning network for the decoder part and discussed the problems encountered in the process of building.