Computer-assisted Quantification of Occlusion and Coil Densities on Angiographic and Histological Images of Experimental Aneurysms

Abstract

Occlusion rates (OR) and coil densities were quantified by computer-assisted morphometry on angiograms and histological ground sections of coil-embolized experimental aneurysms. The aims of this study were 1) to develop computer-assisted evaluations of angiographic OR and histometrical OR, 2) to compare these results to subjectively estimated angiographic OR from clinical practice, and 3) to test the correlation between histometrical data of coil density and occlusion. Eight rabbit carotid-bifurcation aneurysms had been followed by digital subtraction angiography (DSA) before and after Guglielmi detachable coil embolization and at sacrifice (1 h to 24 wk postembolization). Angiographic OR was subjectively estimated, then determined by computer-assisted density-gradient distinction on digitized DSAs. Histometrically, maximum length, neck width, total area, recanalized area, and coil-occupied area were measured on digitized and calibrated color micrographs from surface-stained histological ground sections of the aneurysms. Histometrical OR and coil density were calculated as indirect parameters. Subjective versus computer-assisted angiographic OR yielded for one aneurysm, 100% versus 100%, and for three aneurysms less than 90% versus 65 to 60% occlusion. For four aneurysms, OR was estimated greater than 90%, whereas computer-assisted OR ranged between 45 to 80%, the latter being more precise because of better definition of the aneurysm's total area on digitized DSA. Histometrical OR ranged between 32.8 and 87.6%, but did not correlate significantly with computer-assisted angiographic OR (r = 0.467, P > 0.1) because of differences in two aneurysms. Coil densities between 5.5 and 22.1% were slightly lower than reported in literature but significantly correlated to histometrical OR (r = 0.646, P < 0.05). Computer-assisted DSA evaluation, delivering more precise values than subjectively estimated occlusion, may be a useful tool for follow-up studies. Comparing computer-assisted angiographic with histometrical occlusion demonstrates limits of DSA in displaying the real morphology of coil-embolized aneurysms. The clinically postulated correlation of OR and coil densities was statistically corroborated.