Quantitative analysis of feature pixels in the region of interest of ultrasound images for detection of thermal lesion induced by focused ultrasound

Abstract

The appearance of hyperecho is clinically considered as an indicator of tissue necrosis for ultrasonography-guided high-intensity focused ultrasound (USgHIFU). Nevertheless, there is unobservable hyperechoic change in the absence of cavitation or boiling during sonication. A quantitative analysis that could identify post-sonication change in the ultrasound image is proposed for the detection of HIFU-induced thermal lesion. Before analysis, the acquired images are preprocessed for cancelling of sonication-induced interference, and a region of interest (ROI) is set to cover the focal spot. In the analysis, two parameters that quantify the changes in grayscale of feature pixels in the ROI are used to detect the thermal lesion induced in pork ex vivo. It's found that both parameters can identify more true positives but fewer false negatives than visualizing hyperecho. This study demonstrates the analysis in combination with diagnostic ultrasound system can be used to predict thermal lesion during HIFU treatment.