Anatomy-Guided Representation Learning Using a Transformer-Based Network for Thyroid Nodule Segmentation in Ultrasound Images

TL;DR

This paper introduces SSMT-Net, a semi-supervised transformer-based model that improves thyroid nodule segmentation in ultrasound images by effectively utilizing unlabeled data and multi-task learning to handle ambiguous boundaries and data scarcity.

Contribution

The paper presents a novel semi-supervised, multi-task transformer network that enhances feature extraction and segmentation accuracy in ultrasound thyroid images, addressing data scarcity and boundary ambiguity.

Findings

Outperforms state-of-the-art methods on TN3K and DDTI datasets.

Achieves higher accuracy and robustness in thyroid nodule segmentation.

Effectively leverages unlabeled data for improved feature learning.

Abstract

Accurate thyroid nodule segmentation in ultrasound images is critical for diagnosis and treatment planning. However, ambiguous boundaries between nodules and surrounding tissues, size variations, and the scarcity of annotated ultrasound data pose significant challenges for automated segmentation. Existing deep learning models struggle to incorporate contextual information from the thyroid gland and generalize effectively across diverse cases. To address these challenges, we propose SSMT-Net, a Semi-Supervised Multi-Task Transformer-based Network that leverages unlabeled data to enhance Transformer-centric encoder feature extraction capability in an initial unsupervised phase. In the supervised phase, the model jointly optimizes nodule segmentation, gland segmentation, and nodule size estimation, integrating both local and global contextual features. Extensive evaluations on the TN3K and DDTI datasets demonstrate that SSMT-Net outperforms state-of-the-art methods, with higher accuracy and robustness, indicating its potential for real-world clinical applications.