Aorta detection on endosonographic image sequence: a Hough Transform based architecture

Presents a method to detect the aorta section from an ultrasound image sequence of the esophagus area. The proposed architecture uses the Hough Transform (HT) in order to integrate a priori knowledge. Two primary models compose the algorithm: a fuzzy model, based on data fusion of three different information sources extracted from slices, and a HT model which is often employed for pattern recognition. A global fusion model combines their complementary aspects and advantages. Along the sequence, spatial aorta matching is achieved by parameters propagation and controlled using a 3D trajectory model. Results obtained on a real sequence are presented.