A fully automated approach to aortic distensibility quantification from fetal ultrasound images

Intrauterine growth restriction (IUGR) is a fetal condition that can be assessed by estimating aortic intimamedia thickness (aIMT) and pulse pressure (PP) from ultrasound (US) image sequences. Correct measurement of these quantities requires the identification of the aortic lumen contours at end-systolic (ES) and end-diastolic (ED) phases and the estimation of the undergoing change in diameter (aortic distensibility, ΔD). This analysis currently relies on tedious and error-prone manual tracing. Accordingly, we developed a fully-automated technique for lumen identification and segmentation, allowing direct aortic distensibility estimation, and tested it against manual analysis. The technique is based on convolution with a set of discriminative kernels learned from a training dataset, followed by segmentation based on anisotropic filtering and level-set methods. We tested this approach against manual analysis on 10 image sequences acquired from different subjects, and compared automatically and manually extracted lumen diameter values as well as ΔD values. Results suggest that the proposed technique is as accurate as manual analysis, and could thus serve as a basis for fully-automated aIMT and PP estimation.