Segmentation of plaques in sequences of ultrasonic B-mode images of carotid arteries based on motion estimation and Nakagami distributions

The goal of this work was to perform a segmentation of atherosclerotic plaques in view of evaluating its burden and to provide boundaries for computing properties such as the plaque elasticity distribution (elastogram). The echogenicity of a region of interest comprising the plaque, the lumen, and the adventitia in an ultrasonic B-mode image was modeled by a mixture of three Nakagami distributions. To each of these three tissues corresponds a specific weighted sum of these three distributions, which yields the likelihood of a Bayesian model. The prior of that model includes a geometrical smoothness constraint, as well as an original spatio-temporal cohesion constraint, based on the estimation of the motion field of the plaque in the video sequence. The Maximum A Posteriori (MAP) of the proposed model was computed with a variant of the Exploration/Selection (ES) algorithm. The starting point is a manual segmentation of the first frame. The main contribution of this paper is the estimation of the motion field and its integration into the prior of the Bayesian model. The proposed method was quantitatively compared with manual segmentations of all frames by an expert technician. Various measures were used for this evaluation, including the sensitivity and specificity of the receiver operating characteristic (ROC) analysis, and the error of area of the plaque. Results were evaluated on 21 sequences of 18 symptomatic patients (for a total of 2078 images). We report a sensitivity of 87.7 ± 5.9%, a specificity of 94.6 ± 3.2%, a kappa index of 88.1 ± 4.7%, an error of the plaque area of -0.79 ± 8.0%, and an absolute error of the plaque area of 6.61 ± 4.8%, which compare favorably with previous studies.