Domain-adaptive cross-modal medical image segmentation method, device, equipment and medium based on boundary comparison

By adopting a two-way cross-domain cutmix and dynamic weight allocation strategy based on boundary contrast in medical image segmentation, the problem of poor segmentation performance in cross-modal medical image segmentation is solved, and the discrimination ability and overall segmentation performance of boundary areas are significantly improved.

CN119295491BActive Publication Date: 2025-03-25CARDIOVASCULAR HOSPITAL AFFILIATED TO XIAMEN UNIV
2 Cites 0 Cited by

Patent Information

Application Number
CN202411804970.7
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-12-10
Publication Date
2025-03-25
Estimated Expiration
2044-12-10

AI Technical Summary

Technical Problem

The existing medical image segmentation technology has problems such as network convergence difficulties and easy training collapse in cross-modal medical image segmentation, and ignores object boundaries, resulting in poor segmentation performance.

Method used

The domain adaptation method based on boundary comparison is adopted to construct mixed images and labels through bidirectional cross-domain cutmix, and iterative training is carried out in combination with supervision losses and contrast losses, and the weight of pseudo-labels is dynamically allocated to enhance the model's discriminatory ability in the boundary area.

Benefits of technology

It effectively improves the segmentation performance of cross-modal medical image segmentation, especially at the category boundaries, with low entropy and high confidence, preventing training instability and early performance degradation.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention provides a domain-adaptive cross-modal medical image segmentation method, device, equipment and medium based on boundary contrast, and relates to the technical field of medical image segmentation. The method comprises: obtaining source samples and target domain images from source domain and target domain data sets respectively; using a first image segmentation model as a teacher model to assign weights to pseudo labels of target domain images; obtaining training samples through bidirectional cross-domain cutmix; using a first model with learnable weights as a student model to perform supervision and self-training; combining supervision loss and contrast loss during training to optimize parameters with the goal of pulling query samples toward prototype positive samples and pushing away boundary negative samples, obtaining a second image segmentation model; and finally using the model to segment a second modality medical image. This method has significant advantages in cross-modal medical image segmentation and can effectively improve segmentation performance.
Need to check novelty before this filing date? Find Prior Art