Quantification of the wideband signal associated with bubble collapse is desirable to measure cavitation dosage during microbubble therapy. We have developed a cavitation detection algorithm based on time-frequency analysis with the Stockwell transform. This transform is ideal since it isolates cavitation transients in time and frequency at much higher resolution than other time-frequency methods. In this study, we acquired simultaneous high-speed images and acoustic signals from bubbles undergoing oscillation during acoustic pulses commonly used for permeability enhancement (Transmit: 1.5 MHz, 10-cycle, peak negative pressure range 150–500 kPa, Receive: 15 MHz cavitation detector). Above 200 kPa, linescan vs time images showed relative bubble expansion greater than 2 and echoes contained a wideband signal that coincided with bubble collapse. The Stockwell transform of detected echoes revealed wideband spectrums localized to the time of bubble collapse. After filtering below 4 MHz to reject linear echoes, the root mean square (RMS) of the spectral amplitude was obtained from the time-frequency decomposition, yielding peaks at cavitation transients. The magnitude of the Fourier transform of the RMS signal at 1.5 MHz provided a sensitive indicator for bubble collapse and differed significantly across pressures where cavitation occurred (p < 0.05,t-test). The proposed measurement provides a sensitive metric for cavitation activity during microbubble therapy.