HD-Former: A hierarchical dependency Transformer for medical image segmentation.

Area of Science:

• Medical Image Analysis
• Computer Vision
• Artificial Intelligence

Background:

• Medical image segmentation is crucial for clinical applications.
• Convolutional Neural Networks (CNNs) and Transformer models are used, but hybrid approaches face challenges with multilevel semantic feature gaps and dependencies.
• Existing methods struggle to capture both local and global contextual information effectively.

Purpose of the Study:

• To propose a novel hierarchical dependency Transformer, HD-Former, for improved medical image segmentation.
• To address limitations of existing methods, specifically multilevel semantic feature gaps and dependencies.
• To enhance the capture of hierarchical dense contextual semantic features locally and globally.

Main Methods:

• Introduced the Compressed Bottleneck (CB) module to enrich shallow features and localize target regions.
• Developed the Dual Cross Attention Transformer (DCAT) module for fusing multilevel features and bridging feature gaps.
• Designed the broad exploration network (BEN) combining convolution and self-attention for hierarchical feature extraction.
• Utilized uncertain multitask edge loss to optimize segmentation edges.

Main Results:

• HD-Former demonstrated superior performance compared to state-of-the-art methods on ISIC, LiTS, Kvasir-SEG, and CVC-ClinicDB datasets.
• Achieved significant improvements in both subjective visual quality and objective evaluation metrics.
• Effectively captured hierarchical dense contextual semantic features, addressing prior limitations.

Conclusions:

• HD-Former represents a significant advancement in medical image segmentation.
• The proposed architecture effectively overcomes multilevel semantic feature gaps and dependency issues.
• This method offers enhanced accuracy and robustness for clinical image analysis tasks.