Computerized modeling based on spiral CT data for noninvasive determination of aortic stent-graft length.

Abstract

To preprocedurally determine the correct length of a nonbifurcated endovascular prosthesis for abdominal aortic aneurysm (AAA) repair using a computerized model. A computer program was implemented to calculate the optimal intraluminal course of nonbifurcated stent-grafts from spiral computed tomographic (CT) images of the aortic lumen reconstructed at 2.5, 5, and 10-mm slice thicknesses. The algorithm was tested using 10 phantoms fabricated from 150-mm-long, 10-mm-diameter copper rods that were bent into shapes mimicking different aortic configurations. Midpoint coordinates and rod diameters were determined from each CT image by 3 independent observers and served as input parameters to the program. The influence of the different CT reconstructions on the calculated lengths and possible observer dependence were assessed using calculated length estimation errors. Spiral CT images from 20 consecutive AAA patients scanned before stent-graft implantation were also processed to evaluate the algorithm under clinical conditions. Length estimation errors of the phantoms depended on the degree of bending as well as on the CT reconstruction slice thickness but were observer independent. Maximum errors were 7% for the 10-mm slices, 3.5% for the 5-mm slices, and 1.2% for a 2.5-mm reconstruction. The mean longitudinal shortening of the aorta due to vessel tortuosity was 9.1% +/- 4.8% among the 20 patients. Based on the results of the phantom study, errors of the calculated stent-graft lengths in patients were estimated to be approximately 1% for a 5-mm CT reconstruction increment and <2% for a 10-mm increment. The proposed algorithm makes it possible to calculate noninvasively the correct length of straight stent-grafts under clinical conditions with a 1% to 2% error.