An optimization framework for high intensity focused ultrasound: From design of device to ultrasound therapy

In this study, the optimization framework to seamlessly integrate simulation process for computer simulations which perform from transducer array designs to ultrasound therapy is proposed. The optimization framework consists of three processes: a) design of a random transducer array, b) design of multi-focus patterns and a sparse array, and c) treatment planning of scanning time and path. In an attempt to tackling time-intensive computations, the 3D acoustic pressure expressed by the Rayleigh-Sommerfeld diffraction integral was numerically calculated with the aid of a distributed computing method. Genetic algorithms which were entirely used from the 1st process to the 3rd process also utilized the distributed computing method to speed up the calculation time. For the thermal dose which can describe the tissue damage, the tissue temperature evolution over time was obtained by solving Pennes' bioheat transfer equation on GPUs using NVIDIA's CUDA environment. The proposed optimization framework based on commercial software Isight 5.6 was implemented on the high performance computing facility. With this framework, a 1017-element spherical-section ultrasound phased array was developed and the high speed and safety of focused ultrasound thermal ablation was finally achieved.