Segmentation of deformable structures remains a challenging task in ultrasound imaging especially in low signal-to-noise ratio applications. In this paper a fully automatic method, dedicated to the luminal contour segmentation in intracoronary ultrasound imaging is introduced. The method is based on an active contour with a priori properties that evolves according to the statistics of the ultrasound texture brightness, determined as being mainly Rayleigh distributed. However, contrary to classical snake-based algorithms, the presented technique neither requires from the user the pre-selection of a region of interest tight around the boundary, nor parameter tuning. This fully automatic character is achieved by an initial contour that is not set, but estimated and thus adapted to each image. Its estimation combines two statistical criteria extracted from the a posteriori probability, function of the contour position. These criteria are the location of the function maximum (or maximum o posteriori estimator) and the first zero-crossing of the function derivative. Then starting from the initial contour, a region of interest is automatically selected and the process iterated until the contour evolution can be ignored. In vivo coronary images from 15 patients, acquired with a 20 MHz central frequency Jomed Invision ultrasound scanner were segmented with the developed method. Automatic contours were compared to those manually drawn by two physicians in terms of mean absolute difference. Results demonstrate that the error between automatic contours and the average of manual ones (0.099 ± 0.032mm) and the inter-expert error (0.097 ± 0.027mm) are similar and of small amplitude.

, , , ,
doi.org/10.1117/12.532849, hdl.handle.net/1765/58679
Medical Imaging 2004 - Ultrasonic Imaging and Signal Processing
Department of Cardiology

Brusseau, E., de Korte, C., Mastik, F., Schaar, J., & van der Steen, T. (2004). Fully automatic contour detection in intravascular ultrasound imaging. Presented at the Medical Imaging 2004 - Ultrasonic Imaging and Signal Processing. doi:10.1117/12.532849