Combining spectral and intensity data to identify regions of cavitation in ultrasound images; application to HIFU

The high power intensities in HTFU often result in bubble production, either through cavitation or boiling, which are believed to be a primary contributor to tissue necrosis Bubbles are associated with the appearance of bright hyperechoic regions in ultrasound B-mode images are observed from the bubbles. As they are often the only changes observed during treatment on tissue, some HIFU protocols rely on hyperechoicity as the indicator of tissue lesions. It would be useful clinically to keep a history of cavitation events during treatment which could be recalled on demand to assist planning. A reliable method of determining not just the presence of cavitation but its location is required for this. This paper introduces an active method of cavitation detection through analyzing spectrally the r.f. signal from an imaging transducer which interrogates the region on a pixel by pixel basis. Particular challenges are due to the limited bandwidth of the transducer and to provide good spatial and spectral resolution. Application to ex vivo pig liver using ARMA modeling on time series r.f. data shows that spectral information can be determined at a spatial resolution of about 1mm. The results of spectral estimation are used together with B-mode segmentation in a statistical algorithm to enhance the overall reliability of automatic cavitation detection. © 2007 IEEE.