Passive acoustic mapping utilizing compressed-domain processing for real-time monitoring of cavitation-enhanced drug delivery

Ultrasound-induced cavitation has been shown to improve the delivery of a range of therapeutics including small-molecule drugs, oncolytic viruses and immunotherapies, potentially enhancing their delivery and reducing their toxicity for treatment of solid tumours. Real-time monitoring methods such as passive acoustic mapping (PAM) may be employed to map cavitation activity during treatment. However, while advances in acquisition hardware facilitate outstanding monitoring capability, transfer, storage and processing of the resulting data remains a challenge. Previously we investigated the use of compressed sensing (CS) techniques to sparsely reconstruct full-rate time-series array data for subsequent conventional PAM. Here we extend this approach to directly perform PAM using the compressed-domain (or CS) data. In our approach the CS data is obtained from projections that jointly compress the full-rate array data. The PAM image is computed directly from the CS data using a sparse matching pursuit algorithm, with a dictionary formed from a discretized model using a reference sensor. Results showing accurate localization of cavitation activity using CS-PAM are obtained with simulated and in-vivo (porcine) measured data but processing only 1% of the full-rate data. Finally, common issues with voxel grid density that affect source localization in CS-PAM and conventional PAM are highlighted.