A thinning algorithm to extract medial lines from 3D medical images

This paper presents an efficient parallel algorithm for thinning elongated 3D binary objects (e.g., bony structures, vessel trees, or airway trees). This algorithm directly extracts medial lines (i.e., without creating medial surface). One iteration step consists of 12 subiterations (instead of 6, which is the usual case) according to the selected 12 deletion directions. Our topology preserving algorithm gives satisfactory results for synthetic data tests and for MR angiography brain studies.