Characterization of ultrasound images of HIFU-induced lesions by extraction of its morphological properties

The purpose of this work is to compute several image processing techniques, applied to US images of HIFU-induced lesions, to characterize the shape, contour, position and the orientation of thermal lesions. In order to obtain real-time monitoring of HIFU treatment, a B-mode US imaging system and HIFU were synchronized. First, the HIFU sonication procedure consisted of applying an initial power of 10 W over an 8% BSA tissue-mimicking polyacrylamide gel until a hyperechoic spot appeared. Afterwards, the power was gradually increased up to 40 W for 4 min. The frame acquisition rate was set to 5 fps. Then, the region of interest (ROI) was manually selected and an image segmentation algorithm was implemented to isolate the thermal lesion based on different image pre-processing techniques and morphological operations. Thereafter, parameters such as area growing, the center of mass, eccentricity and the orientation of the equivalent ellipse were calculated to extract some morphological lesion features. This approach can be used as feedback system of the thermal lesion formation to determine more accurately the ablation zone.