Share Email Print

Proceedings Paper

Closing of interrupted vascular segmentations: an automatic approach based on shortest paths and level sets
Author(s): Nils Daniel Forkert; Alexander Schmidt-Richberg; Dennis Säring; Till Illies; Jens Fiehler; Heinz Handels
Format Member Price Non-Member Price
PDF $14.40 $18.00
cover GOOD NEWS! Your organization subscribes to the SPIE Digital Library. You may be able to download this paper for free. Check Access

Paper Abstract

Exact segmentations of the cerebrovascular system are the basis for several medical applications, like preoperation planning, postoperative monitoring and medical research. Several automatic methods for the extraction of the vascular system have been proposed. These automatic approaches suffer from several problems. One of the major problems are interruptions in the vascular segmentation, especially in case of small vessels represented by low intensities. These breaks are problematic for the outcome of several applications e.g. FEM-simulations and quantitative vessel analysis. In this paper we propose an automatic post-processing method to connect broken vessel segmentations. The approach proposed consists of four steps. Based on an existing vessel segmentation the 3D-skeleton is computed first and used to detect the dead ends of the segmentation. In a following step possible connections between these dead ends are computed using a graph based approach based on the vesselness parameter image. After a consistency check is performed, the detected paths are used to obtain the final segmentation using a level set approach. The method proposed was validated using a synthetic dataset as well as two clinical datasets. The evaluation of the results yielded by the method proposed based on two Time-of-Flight MRA datasets showed that in mean 45 connections between dead ends per dataset were found. A quantitative comparison with semi-automatic segmentations by medical experts using the Dice coefficient revealed that a mean improvement of 0.0229 per dataset was achieved. In summary the approach presented can considerably improve the accuracy of vascular segmentations needed for following analysis steps.

Paper Details

Date Published: 12 March 2010
PDF: 8 pages
Proc. SPIE 7623, Medical Imaging 2010: Image Processing, 76233G (12 March 2010); doi: 10.1117/12.844210
Show Author Affiliations
Nils Daniel Forkert, Univ. Medical Ctr. Hamburg-Eppendorf (Germany)
Alexander Schmidt-Richberg, Univ. Medical Ctr. Hamburg-Eppendorf (Germany)
Dennis Säring, Univ. Medical Ctr. Hamburg-Eppendorf (Germany)
Till Illies, Univ. Medical Ctr. Hamburg-Eppendorf (Germany)
Jens Fiehler, Univ. Medical Ctr. Hamburg-Eppendorf (Germany)
Heinz Handels, Univ. Medical Ctr. Hamburg-Eppendorf (Germany)

Published in SPIE Proceedings Vol. 7623:
Medical Imaging 2010: Image Processing
Benoit M. Dawant; David R. Haynor, Editor(s)

© SPIE. Terms of Use
Back to Top