Prediction and suppression of HIFU-induced vessel rupture using passive cavitation detection in an ex vivo model

BACKGROUND: Occlusion of blood vessels using high-intensity focused ultrasound (HIFU) is a potential treatment for arteriovenous malformations and other neurovascular disorders. However, attempting HIFU-induced vessel occlusion can also cause vessel rupture, resulting in hemorrhage. Possible rupture mechanisms include mechanical effects of acoustic cavitation and heating of the vessel wall. METHODS: HIFU exposures were performed on 18 ex vivo porcine femoral arteries with simultaneous passive cavitation detection. Vessels were insonified by a 3.3-MHz focused source with spatial-peak, temporal-peak focal intensity of 15,690–24,430 W/cm(2) (peak negative-pressure range 10.92–12.52 MPa) and a 50% duty cycle for durations up to 5 min. Time-dependent acoustic emissions were recorded by an unfocused passive cavitation detector and quantified within low-frequency (10–30 kHz), broadband (0.3–1.1 MHz), and subharmonic (1.65 MHz) bands. Vessel rupture was detected by inline metering of saline flow, recorded throughout each treatment. Recorded emissions were grouped into ‘pre-rupture’ (0–10 s prior to measured point of vessel rupture) and ‘intact-vessel’ (>10 s prior to measured point of vessel rupture) emissions. Receiver operating characteristic curve analysis was used to assess the ability of emissions within each frequency band to predict vessel rupture. Based on these measurements associating acoustic emissions with vessel rupture, a real-time feedback control module was implemented to monitor acoustic emissions during HIFU treatment and adjust the ultrasound intensity, with the goal of maximizing acoustic power delivered to the vessel while avoiding rupture. This feedback control approach was tested on 10 paired HIFU exposures of porcine femoral and subclavian arteries, in which the focal intensity was stepwise increased from 9,117 W/cm(2) spatial-peak temporal-peak (SPTP) to a maximum of 21,980 W/cm(2), with power modulated based on the measured subharmonic emission amplitude. Time to rupture was compared between these feedback-controlled trials and paired controller-inactive trials using a paired Wilcoxon signed-rank test. RESULTS: Subharmonic emissions were found to be the most predictive of vessel rupture (areas under the receiver operating characteristic curve (AUROC) = 0.757, p < 10(-16)) compared to low-frequency (AUROC = 0.657, p < 10(-11)) and broadband (AUROC = 0.729, p < 10(-16)) emissions. An independent-sample t test comparing pre-rupture to intact-vessel emissions revealed a statistically significant difference between the two groups for broadband and subharmonic emissions (p < 10(-3)), but not for low-frequency emissions (p = 0.058). In a one-sided paired Wilcoxon signed-rank test, activation of the control module was shown to increase the time to vessel rupture (T(-) = 8, p = 0.0244, N = 10). In one-sided paired t tests, activation of the control module was shown to cause no significant difference in time-averaged focal intensity (t = 0.362, p = 0.363, N = 10), but was shown to cause delivery of significantly greater total acoustic energy (t = 2.037, p = 0.0361, N = 10). CONCLUSIONS: These results suggest that acoustic cavitation plays an important role in HIFU-induced vessel rupture. In HIFU treatments for vessel occlusion, passive monitoring of acoustic emissions may be useful in avoiding hemorrhage due to vessel rupture, as shown in the rupture suppression experiments.