Semi-TMS: an efficient regularization-oriented triple-teacher semi-supervised medical image segmentation model.

Area of Science:

• Medical Image Analysis
• Artificial Intelligence in Healthcare
• Computer Vision

Background:

• Medical image segmentation is crucial for diagnosis but requires extensive labeled data.
• Convolutional Neural Networks (CNNs) and Transformers need large datasets, which are costly and time-consuming to annotate.
• Semi-supervised learning offers a solution by utilizing limited labeled data with abundant unlabeled data.

Purpose of the Study:

• To enhance medical image segmentation performance using a novel semi-supervised approach.
• To leverage unlabeled data effectively through a triple-teacher cross-learning framework.
• To incorporate shape perception and multi-scale consistency regularization for improved accuracy.

Main Methods:

• Proposed Semi-TMS: a multi-scale semi-supervised method with three-teacher cross-learning and shape perception.
• Utilized a hybrid model architecture with two CNN teachers (A, C) and one Transformer teacher (B).
• Implemented cross-learning between CNNs and Transformer, incorporating shape perception and multi-scale consistency regularization.

Main Results:

• The proposed Semi-TMS method outperformed existing semi-supervised segmentation models on public datasets.
• Demonstrated effective utilization and improved accuracy of unlabeled data via multi-scale consistency.
• Implicitly captured shape information, enhancing segmentation robustness.

Conclusions:

• Semi-TMS effectively improves medical image segmentation accuracy with limited labeled data.
• The method shows potential for practical application in clinical diagnosis and research.
• It offers a valuable auxiliary tool for clinicians and researchers utilizing medical imaging.