Ex-vivo 3D assessment of carotid stenosis with ultrasound

Abstract

Atherosclerosis causes heterogeneous remodeling of arterial structure and composition in the carotid vessel wall. It has been shown that the progression of the disease can be monitored by tracking changes in the carotid intima-media thickness (IMT). Non-invasive peripheral vascular ultrasound (U/S) of the carotid artery is a non-invasive, cost effective, accepted means of measuring IMT. Traditionally, evaluation of IMT in the carotid has been limited to 2D U/S scans. This method is disadvantageous as 2D scans are scan plane dependent, limiting the area over which one can evaluate the extent of the disease. Reproducing the identical scan plane on subsequent scans is also difficult. Evaluation of the carotid vessel wall in 3D will allow for a more complete and reproducible assessment of disease through IMT measurements. We have constructed a fully 3D image processing scheme for analyzing carotid U/S volumes to extract the inner and outer vessel wall boundaries. Sequences of 2D B-mode U/S cross sections of ex-vivo carotid specimens are collected and voxelized to create 3D U/S volumes. By applying a 3D directionally sensitive, edge preserving filter to the U/S volumes, we obtain 3D edge fields that are more distinct than traditional gradient edge fields. Initial point selection of the boundaries, together with these enhanced 3D edge fields, are used with a deformable surface to extract the final inner and outer vessel boundaries. Through intra- and inter-observer tests on IMT differences, we show that the 3D boundaries extracted using our automatic technique are more reproducible than boundaries extracted from manual point selection.