Algorithms for Improved Image Guidance and Automatically Optimized Treatment Planning for C-arm Cone Beam CT Guided Histotripsy - PROJECT SUMMARY Primary liver cancer is a leading cause of cancer-related deaths worldwide with incidence rates rapidly increasing. Histotripsy is a focal tumor therapy that has been clinically adopted with enthusiasm since being approved by the FDA in October 2023 for treating liver tumors under diagnostic ultrasound (US) guidance. It is a noninvasive, nonionizing, and nonthermal focused US therapy that ruptures cells mechanically through acoustic cavitation, offering precise and controlled tumor destruction. Despite its therapeutic potential, its clinical workflow is rudimentary consisting of undesirably long procedures and often inadequate imaging for visualizing and targeting tumors. C-arm cone beam CT (CBCT) guided histotripsy is being developed to overcome US visualization limitations. However, CBCT-guided histotripsy still shares many of the same challenges as US- guided treatments unless the workflow is significantly improved. To enable histotripsy treatment delivery, a large, deformable acoustically coupling water bath (10-14 L) is placed on the patient’s abdomen and requires iterative, time-consuming positioning. Treatment planning and tumor targeting are both currently performed intraprocedurally, restricting time, limiting accuracy, and further increasing the procedure duration. Treatments are delivered without motion compensation, destroying surrounding nontarget tissue. This proposal aims to address these challenges by developing a patient-specific and automated treatment planning workflow for C- arm CBCT guided histotripsy, enabling efficient and accurate histotripsy treatments. To this end, the optimal water bath and transducer positions will be predetermined by modeling abdominal deformations from the water bath. Treatment planning in advance will be enabled by registering high quality pre-procedural imaging depicting the target tumor, surrounding anatomy, and related planning parameters to intraprocedural imaging, accounting for patient positioning. To compensate for respiratory motion, the prescribed treatment will be adjusted based on predicted respiratory-induced deformations. All of such will be developed using fully-automated, patient- specific approaches through finite element (FE) modeling and deep learning. In doing so, I will learn how to employ deep learning networks, implement physics-based deformation models, lead animal experiments and a human subject study, improve my communication skills, develop a strong network, and cultivate ethical research practices. I will train at the University of Wisconsin-Madison, the home of the largest Medical Physics doctoral program in the world and of an innovative Radiology department. I will be mentored by Dr. Martin Wagner and Dr. Paul Laeseke (technical and clinical experts of CBCT guided histotripsy and image guided therapy) and other experts in X-ray physics, FE modeling, and machine learning. Industry leaders in imaging and histotripsy will provide devices/support to facilitate this work. In accomplishing the proposed aims and training, I will utilize a C- arm to improve the treatment planning workflow for histotripsy, and gain the invaluable skills to become a scientific leader in image guided therapy.
