Feasibility of a Curvature-based Enhanced Display System for Detecting Cerebral Aneurysms in MR Angiography

Abstract

To assess the feasibility of a curvature-based enhanced display system for detecting cerebral aneurysms in MR angiography. MR angiography studies of 18 patients (eight male and 10 female, average age 65.7, age range 50 to 75 years old) with 23 known aneurysms were evaluated with a curvature-based display system. The two curvature features-the volumetric shape index and curvedness values-were calculated at each voxel. These were displayed independently on a workstation, overlaid on volume-rendered images. Two neuroradiologists evaluated the images for visibility and diagnosis of the cerebral aneurysms. The diagnostic results were compared with the original reports. The calculation time for each curvature index was 30 to 40 s for 120 to 140 slices of original MR angiography data. Shape index images emphasized smooth and round aneurysms more than aneurysms with irregular surfaces. Curvedness images revealed aneurysms well when the aneurysms had diameters that differed from those of the surrounding vessels. The computer-assisted-detection method detected 24 aneurysms, three of which were not pointed out in the initial report. Our results show that the curvature-based display system we have developed is feasible and that it may help to detect small aneurysms that are prone to be overlooked in routine readings.