Interventional Cardiology and Angiology

• The intravascular coronary ultrasound (ICUS) permits cross-sectional tomographic imaging of coronary arteries and provides detailed insight into the pathology of coronary artery disease. Although the detection of the endovascular wall and the assessment of geometrical parameters in conventionally obtained ICUS images is available by semi-automated systems, there are several limitations to their application. To date, contour detection systems for 3-D reconstruction of the endovascular wall are based on the "volume rendering" algorithm and allow interactive manual tracing by using planar analysis for the quantification of geometrical structures in selected single images. Manual contour tracing even in so called automated systems is subjective, time-consuming and hampers the additionally yielded information.
• We therefore developed an automatic contour detection system of the endovascular surface by using a new area orientated contour detection for 3-D visualization of the vessel wall based on the "surface modeling" algorithm.
• The assessment of geometrical vessel parameters can be achieved without additional manual tracing. We examined 10 coronary arteries in 10 patients with stent implantation due to coronary artery disease with a 2.9 f /30 Mhz ICUS catheter (System CVIS) to evaluate accuracy of the automated wall detection and the calculation of diameter, area and volume of the reconstructed lumen surface. Image acquisition was performed with an automatical pullback device (1mm/s). The digitalization of the obtained images was realized on equidistant time points (delta t=1s). The validation of the wall detection was compared with manual tracing.The automated contour detection revealed a high correlation of internal vessel surface and measured diameter and area stenosis compared to manual tracing. Intraobserver and interobserver accuracy in area measurements of lumen and area stenosis were 93% (-4.1%+9.3%) and 94% (-6.3%+11%).
• These results indicate that our automated contour algorithm provides internal vessel surface detection with a high accuracy and permits reproducible geometrical measurements in atherosclerotic arteries.