Evaluation of 3D reconstructed lower limb vessel geometries with an ultrasound robotic imaging system

Abstract

The localization of lesions, their lengths and the degree of stenosis are the most common criteria used to assess the severity of lower limb peripheral arterial disease (PAD). 2D-ultrasound (US) imaging is the first-line diagnosis method used to investigate arterial lesions; however, it cannot render a 3D map of the entire lower limb vascular tree required for therapy planning. Moreover, current 3D-US developments are not optimally adapted for this particular clinical application. We proposed a prototype 3D-US imaging robotic system that can control and standardize the image acquisition process to reconstruct accurately arteries from the iliac down to the popliteal. Because calibration has a major impact on the quality of reconstructed geometries, a customized Z-phantom calibration procedure was first implemented. At optimum US settings, the calibration transform was evaluated with a reconstruction precision < 1.10 mm. The calibration transform accuracy was also evaluated on two vascular phantoms of lower limb mimicking vessel geometries. Reconstruction performances were assessed in distance errors and cross-sectional areas. The mean reconstruction distance error was 0.39 ± 0.35 mm for the axisymmetric cylindrical phantom, whereas it was 1.38 ± 1.29 mm for the realistic reproduction of a diseased iliac artery. Similar findings were found for the area measures. Altogether, these results demonstrate the potential of the robot to represent adequately lower limb vessels for the clinical evaluation of stenoses.