Trans-MT: a 3D semi-supervised glioma segmentation model integrating transformer architecture and asymmetric data augmentation

Key Points

• The segmentation model achieved significant accuracy in identifying glioma boundaries in medical imaging data, enhancing diagnosis accuracy.
• Key metric shows an improved performance score of 92% across diverse datasets, indicating its robustness.
• Application of a 3D semi-supervised learning approach combined with transformer architecture offers novel insights into glioma segmentation effectiveness.
• Implications may enable clinicians to better visualize and assess tumor boundaries, improving treatment planning.