The potential of time-multiplexed steering in phased array microwave hyperthermia for head and neck cancer treatment.

Clinical studies have shown that hyperthermia sensitizes tumor cells for conventional therapies. During phased-array microwave hyperthermia, an array of antennas is used to focus the electromagnetic waves at the target region. Selective heating, while preserving the healthy tissue, is a demanding challenge and currently patient specific pre-treatment planning is used to optimize the amplitudes and phases of the waves. In addition, when needed, this single optimal heat distribution is adapted using the simulations based on the feedback from thermo-sensors and the patient. In this paper, we hypothesize that sequential, i.e. 'time-multiplexed', application of multiple Pareto optimal heating patterns provides a better time-averaged treatment quality. To test the benefit of such a time-multiplexed approach, a multi-objective genetic algorithm was introduced to balance two objectives that both focus the specific absorption rate (SAR) delivered to the target region but differ in the suppressing of pre-defined hotspots. This step leads to two Pareto optimal distributions. These 'diverse' antenna settings are then applied sequentially and thermal simulations are used to evaluate the effectiveness of the time-multiplexed steering. The proposed technique is tested using treatment planning data of a representative dataset of five head and neck patients for the HYPERcollar3D. Steering dynamics are analysed and the time-multiplexed steering is compared to the current static solution used in the clinic, i.e. hotspot-target SAR quotient optimization using particle swarm optimization. Our results demonstrate that realistic steering periods of 10s suffice to stabilize temperatures within 0.04 °C and the ability to enhance target heating while reducing hotspots, i.e. 0.3 °C-1.2 °C improvement in T while reducing hotspot temperatures by 0.6 °C-1.5 °C.