DBE-Net: Dual Boundary-Guided Attention Exploration Network for Polyp Segmentation.

Area of Science:

• Medical Imaging
• Computer Vision
• Artificial Intelligence

Background:

• Accurate polyp segmentation during colonoscopy is crucial for early cancer detection and treatment.
• Existing methods struggle with blurry polyp boundaries, multi-scale variations, and distinguishing polyps from normal tissues.

Purpose of the Study:

• To develop an advanced deep learning network for precise automatic polyp segmentation.
• To address the limitations of current polyp segmentation techniques, improving diagnostic accuracy.

Main Methods:

• Proposed a dual boundary-guided attention exploration network (DBE-Net) for polyp segmentation.
• Introduced a dual boundary-guided attention module using a coarse-to-fine strategy for boundary refinement.
• Incorporated a multi-scale context aggregation enhancement module and a low-level detail enhancement module.

Main Results:

• DBE-Net demonstrated superior performance and generalization across five benchmark datasets.
• Achieved excellent mDice scores of 82.4% on CVC-ColonDB and 80.6% on ETIS.
• Outperformed state-of-the-art methods by 5.1% and 5.9% on these challenging datasets, respectively.

Conclusions:

• DBE-Net effectively addresses key challenges in polyp segmentation, including boundary ambiguity and scale variation.
• The proposed network offers a significant advancement in automated polyp detection for colonoscopy.
• This technology has the potential to improve early cancer detection rates and patient outcomes.