VesselDigitizer: A deep learning-powered 3D Slicer extension for digitizing cerebral vasculature from 3D medical images

Motivation: Accurate and fast digitization of the cerebral vasculature can facilitate research on cerebrovascular health, the development of novel imaging biomarkers, and clinical applications. Goal(s): To develop a user-friendly tool with high flexibility and extensibility to segment, label, and quantify cerebral vasculature by taking advantage of deep learning. Approach: The tool VesselDigitizer was developed as an extension of 3D Slicer to leverage its image processing and visualization capability. Deep-learning powered modules for segmentation, refinement, and labeling of the vasculature were developed. Results: VesselDigitizer enables fast segmentation and vessel labeling and provides various handy tools to identify and correct incorrect segmentation and labeling. Impact: VesselDigitizer makes large-scale analysis of cerebral vasculatures easier and more accurate.