Kenneth Richard Hoffmann PhD, Presenter: Nothing to Disclose

Alan M. Walczak MS, Abstract Co-Author: Nothing to Disclose

Peter B. Noël, Abstract Co-Author: Nothing to Disclose

Petru Dinu, Abstract Co-Author: Nothing to Disclose

Ricardo Hanel, Abstract Co-Author: Nothing to Disclose

Jacek Dmochowski PhD, Abstract Co-Author: Nothing to Disclose

Lee R Guterman MD, Abstract Co-Author: Nothing to Disclose

L. Nelson Hopkins, Abstract Co-Author: Nothing to Disclose

et al, Abstract Co-Author: Nothing to Disclose

Three-dimensional presentation of carotid arteries and of vessel curvature and tortuousity may facilitate neuro-endovascular procedures, specifically by aiding decisions as to whether a device can access the intervention site. Therefore, we have developed methods to reconstruct 3D vessels and calculate tortuousity measures using two or more projections acquired during clinical procedures.

We routinely analyze clinical cases in which two or more projection views of the access vessels (carotid or vertebral) are obtained. For each view, images with opacified arteries are selected, and the vessel centerline points are indicated. Indicated centerlines are fit using cubic splines. The relative imaging geometries are first estimated using the gantry information. Epipolar lines are generated for each endpoint and projected into the other image. The geometries are iteratively adjusted until the two pairs of endpoints align with their corresponding epipolar lines. The 3D centerlines are generated using the determined imaging geometry and epipolar constraints. When more than two views are available, the geometries are iteratively refined until differences between corresponding 3D centerline points are minimized, and the centerlines are averaged to yield the final centerline. The vessel size is determined for each centerline point, and the vessel is rendered for viewing. For each centerline point, curvature and tortuousity are calculated with tortuousity determined as the ratio of the arclength between centerline points separated by a specified Euclidean distance divided by that distance.

Prior to adjustment, poor quality vessel centerlines were often obtained. After adjustment, centerlines are smooth and in good agreement with the 2D projections (errors < 1 mm). Tortuousity and curvature measures are well correlated, with tortuousity values greater than 1.3 corresponding to regions with vessel loops or sharp turns.

We have developed methods for providing smooth, reliable 3D vessel centerline and tortuousity data from two or more angiographic views. Supported by NIH grants RO1 EB002916, R01 EB002873, R01 NS43924, and the Toshiba Corporation.

Hoffmann, K, Walczak, A, Noël, P, Dinu, P, Hanel, R, Dmochowski, J, Guterman, L, Hopkins, L, et al, , Clinical System for 3D Analysis of Access Arteries from Two or More Views.  Radiological Society of North America 2004 Scientific Assembly and Annual Meeting, November 28 - December 3, 2004 ,Chicago IL. http://archive.rsna.org/2004/4412589.html