VESCL: an open source 2D vessel contouring library.

Area of Science:

• Medical Image Analysis
• Computer-Assisted Surgery
• Biomedical Engineering

Background:

• Accurate vessel segmentation is crucial for medical image analysis.
• Manual segmentation is labor-intensive and time-consuming.
• Existing automated methods require high-quality training data.

Purpose of the Study:

• Introduce VESCL, a novel library for computer-assisted 2D vessel contouring and segmentation.
• Facilitate the generation of gold-standard datasets for training and validating automated segmentation algorithms.
• Improve the efficiency and accuracy of manual vessel segmentation.

Main Methods:

• VESCL is an open-source C++ library for medical image processing.
• It offers an intuitive interface for drawing parametric curves along vessels.
• Utilizes optimized filtering for automatic curve fitting to vessel centerlines and width determination.
• Supports export of various segmentation representations (binary, occupancy maps, distance fields).

Main Results:

• Achieves sub-pixel resolution for vessel centerlines and widths.
• Demonstrates a 22x speed increase over conventional manual segmentation and 3x over existing computer-assisted methods.
• Segmentation accuracy is comparable to inter-operator variability for gold-standard datasets.
• Shows promising results in neurovascular DSA, coronary artery DSA, and retinal artery images.

Conclusions:

• VESCL significantly reduces the effort required for manual gold-standard vessel segmentation.
• It serves as a valuable tool for developing and evaluating automated segmentation techniques.
• The library's open-source nature promotes wider adoption and collaboration in medical image analysis.