Coherence-Factor-based Passive Acoustic Mapping for Real-Time Transcranial Cavitation Monitoring with Improved Axial Resolution

Traditional passive acoustic mapping (PAM) based on the time exposure acoustic (TEA) algorithm for focused ultrasound (FUS) brain therapy suffers from a poor axial resolution with a clinically-relevant frequency (< 3MHz) for transcranial imaging. Although some methods have been proposed to achieve better resolution, the real-time implementation is virtually impossible due to their computational complexity. Here, we developed a coherence factor (CF)-based PAM that can enhance the axial resolution with a low computational cost. In the proposed method, a CF across the delayed channel data was used as a weighting factor and the delay-and-sum with CF-weighting was repeated for every time sample and accumulated over the PAM integration time. The method was implemented on a Verasonics Vantage system with a GPU using CUDA. The CF exhibited better resolution than the TEA-PAM and comparable to the eigenspace-based robust Capon beamformer (EISRCB)-based PAM. The computational efficiency of CF-PAM was approximately 100 times higher than EISRCB-PAM, which in return enabled real-time monitoring. In the ex-vivo skull study, the CF-PAM successfully located the cavitation transcranially with improved resolution. While EISRCB required a careful parameter adjustment, the CF-PAM did not need any adjustment and provided reliable results throughout all the frames. In conclusion, this technique could help to monitor cavitation activities in real time during the sonication as part of several therapeutic ultrasound applications.