What question did this study set out to answer?

The aim is to create deep learning models that enhance the accuracy and efficiency of medical image segmentation, addressing data challenges.

March 28, 2026Open Access

Efficient and Generalised Deep Learning Models for Medical Image Segmentation

View Full Paper

Ask AI

Bookmark

Authors

This thesis develops innovative deep learning models to improve medical image segmentation efficiency and generalisation in clinical settings.

The aim is to create deep learning models that enhance the accuracy and efficiency of medical image segmentation, addressing data challenges.
Developed CTranS, combining CNN and Transformer for baseline segmentation.
Introduced MO-CTranS for learning from partially labeled datasets.
Proposed CRFTrans to improve computational efficiency with a novel recursive layer.
Created SSL-MedSAM2, a semi-supervised framework that integrates few-shot learning.
CTranS achieved state-of-the-art performance in medical image segmentation.
MO-CTranS enhanced generalisability and tackled label inconsistencies.
CRFTrans reduced computational complexity without sacrificing performance.
SSL-MedSAM2 significantly decreased the need for manual annotations, improving data efficiency.

Zhendi Gong (2026) studied this question.

View Full Paper

Ask AI

Bookmark

March 28, 2026Open Access

View Full Paper

Ask AI

Bookmark

Authors

This thesis develops innovative deep learning models to improve medical image segmentation efficiency and generalisation in clinical settings.

The aim is to create deep learning models that enhance the accuracy and efficiency of medical image segmentation, addressing data challenges.
Developed CTranS, combining CNN and Transformer for baseline segmentation.
Introduced MO-CTranS for learning from partially labeled datasets.
Proposed CRFTrans to improve computational efficiency with a novel recursive layer.
Created SSL-MedSAM2, a semi-supervised framework that integrates few-shot learning.
CTranS achieved state-of-the-art performance in medical image segmentation.
MO-CTranS enhanced generalisability and tackled label inconsistencies.
CRFTrans reduced computational complexity without sacrificing performance.
SSL-MedSAM2 significantly decreased the need for manual annotations, improving data efficiency.

Zhendi Gong (2026) studied this question.

View Full Paper

Ask AI

Bookmark