Compact magnetic resonance-guided high-intensity focused ultrasound phased array system

Abstract

Phased array ultrasound transducers (Txs) can induce electronically controllable localized hyperthermia to tissue deep within the human body. This localized heating can be applied to tumours as an adjunct therapy in multimodal oncology treatments. Magnetic resonance-guided high-intensity focused ultrasound (MRgHIFU) systems utilize imaging techniques to monitor tissue temperature during sonication, providing feedback for control of the Tx’s beam-forming. The electromagnetic environment within and near the magnetic resonance (MR) bore restricts the design of driving solutions. The arrays are normally driven by linear amplifiers and matching networks placed at a distance from the MR bore. As clinical arrays can contain over 2000 Txs, this method of driving can result in significant costs and heating. This thesis proposes a novel MRgHIFU phased array system integration, and an associated rapid design and production processes. The system’s array is designed for conformal use in 3T MR head and neck hyperthermia applications. The array is powered using a high-efficiency quasi class-DE (qDE) driving method to yield equal acoustic output from all Txs without use of inductors. This allows the drivers to be placed directly within the MR bore. The driver gating signals are controlled by a field-programmable gate array (FPGA) also placed inside the MR bore. A rapid design and production process is proposed to facilitate case-by-case target oriented arrays, which yield smaller arrays that can be used as conformal devices. These conformal transducers can help facilitate treatment of targets in challenging locations, and help compensate for patient movement. A Tx casing design was developed for individual ceramics to facilitate repeated removal from various array shells. These removable Txs can be re-used in multiple arrays as they are tuned for qDE driving independent of the array geometry A series of MATLAB functions were created to find sparse array designs by running thousands 3D k-space pseudospectral heterogeneous acoustic simulations, varying the pseudo-random Tx positions, array geometry, or steering position between each simulation. The designed functions are used to sweep array geometry and randomized element positions to find a configuration that yields sufficient performance. Once an acoustically correct solution is found, it is verified by executing k-Wave thermal diffusion simulations based on the acoustic simulation’s resultant pressure waves. Heating was applied for 30 minutes using a discrete PID controller based on temperature feedback. The array geometry can be exported from MATLAB for use in Autodesk Inventor to 3D print the array exactly as represented in simulation. Multiple groups of Txs were electrically characterized then binned until an appropriate qDE driving solution was achievable.