ACE-Net: A-line coordinates encoding network for vascular structure segmentation in ultrasound images.

Area of Science:

• Medical Imaging
• Biomedical Engineering
• Computer Vision

Background:

• Ultrasound (US) imaging is crucial for real-time vascular evaluation and guiding procedures.
• Accurate segmentation of vascular structures aids in precise localization and measurement.
• Current US segmentation methods often require post-processing for contour extraction.

Purpose of the Study:

• To introduce ACE-Net, a novel deep learning approach for direct contour coordinate prediction in ultrasound images.
• To enhance the efficiency and accuracy of vascular structure segmentation during US-guided interventions.

Main Methods:

• ACE-Net employs a dual-module architecture: boundary regression for coordinate prediction and A-line classification for target area identification.
• The method directly predicts contour coordinates for each ultrasound scanning line (A-line).
• Evaluation was performed on three clinical ultrasound datasets using metrics like Dice Similarity Coefficient (DSC) and inference time.

Main Results:

• ACE-Net demonstrated superior inference time compared to state-of-the-art segmentation methods.
• The proposed method achieved comparable or superior Dice Similarity Coefficient (DSC) performance.
• Direct contour prediction eliminated the need for post-processing steps.

Conclusions:

• ACE-Net offers an efficient and accurate solution for vascular structure segmentation in ultrasound imaging.
• The direct contour prediction approach streamlines the segmentation workflow.
• The method shows significant potential for improving clinical applications of ultrasound image analysis.