Automated HIFU treatment planning and execution based on 3D modeling of the prostate, urethra, and rectal wall

Current planning of prostate cancer treatment with an image-guided high-intensity focused ultrasound (HIFU) device consists of defining treatment zones on up to 15 ultrasound images that span the entire prostate. While effective, this process may be time-consuming in the case of large prostates or if multiple treatments are required. For this reason, a method has been developed to model quickly the prostate, urethra, and rectal wall from 2D ultrasound images to allow for completely automated HIFU treatment planning. It is based on the identification of boundaries and prostate structures in 4 to 6 orthographic ultrasound images prior to treatment. The structures are defined through traces and form the input to structure-specific 3D models. All structures are modeled as deformable parametric surfaces that are then processed by the automated treatment planning module. Early results show that the developed models are robust, flexible, generate good closed surfaces even in the absence of full imaging data, and accurately model the wide variation encountered in tracing and pointing skills, human prostates, urethras, and rectal wall shapes. The model accuracy is sufficient for planning these HIFU treatments. The entire treatment planning process is shown, highlighting the usefulness of the developed 3D models. Specific model details for the prostate, urethra, and rectal wall are given. Results using in-vivo animal and human data as model inputs and envisioned treatment planner outputs are also presented, together with the initial activities for integration into the HIFU device.