Meeting Banner
Abstract #2565

Towards Robust & Fast Vessel Extraction from MRA Images

Maysa M. Garcia Macedo1, Choukri Mekkaoui2, Marcel Parolin Jackowski1

1Computer Science, University of So Paulo, So Paulo, SP, Brazil; 2Radiology, Harvard Medical School, Boston, MA, United States


Current techniques for the analysis of vascular networks are time-consuming and limited in their ability to assess the integrity and functional state of vascular architectures. In this work, we evaluate a novel technique based on the well-known Hough transform to dynamically estimate the trajectory and vessel diameter without resorting to time-consuming multiscale techniques. Results are validated using synthetic datasets and applied on a magnetic resonance angiography (MRA) image of a human brain. The proposed technique yielded high reproducibility rates, robustness to different noise levels, associated with simplicity of execution.