Double-Difference Framework for Improved Localization of Cavitation Events in Passive Cavitation

Cavitation under ultrasound enables powerful therapeutic effects, but accurate localization remains a major challenge due to poor resolution and noise sensitivity in conventional imaging. We present a double-difference passive cavitation mapping (DPCM) framework, adapted from seismology, that leverages relative arrival times and incorporates focal and self-consistency priors for robust localization. The method was validated in a wire phantom, a vessel phantom with microbubbles, and an in vivo tumor-bearing mouse model. Across all settings, DPCM outperformed standard approaches, reducing localization error and variability while aligning cavitation activity with physical boundaries. In particular, DPCM achieved 85.3% in-region localization in vessel phantoms and 73.5% in tumors, demonstrating reliable targeting even in complex biological environments. These results establish DPCM as a robust and accurate framework for real-time cavitation monitoring, offering strong potential to guide therapeutic ultrasound with sub-millimeter precision.