Kavitha Arunachalam

The performance of the genetic algorithm (GA) and particle swarm optimization (PSO) was compared to identify the best-suited algorithm for hyperthermia treatment planning (HTP) of breast cancer. Both algorithms were tested on four heterogeneous patient breast models derived from magnetic resonance (MR) images. Electromagnetic (EM) simulations indicate that PSO induces 5.7% less hotspot to target quotient (HTQ) compared to GA. However, coupled EM and thermal simulations of four patient models indicate that GA based HTP induces 1.25 °C - 3.87 °C higher average temperature in cancer tissue with limited thermal hotspots in healthy tissue when compared to PSO algorithm. This was observed to be due to the low power level assigned to each channel by PSO compared to GA. Coupled simulations of heterogeneous patient models indicate GA is a better global optimization algorithm for HTP of breast cancer.

Genetic algorithm-based array thinning technique is investigated in this paper to study the performance of our 18-element 434 MHz phased array breast applicator when driven by reduced number of active antennas. Selective power deposition ability was assessed for 18, 15, 12, and 9 active antennas on 25 patient models with varying characteristics. The average hotspot to target quotient of 18-, 15-, 12-, and 9-antenna excitation in 25 patient models was 1.18, 1.09, 1.09, and 1.13, respectively. The average temperature in 50% tumor volume for 18-, 15-, 12-, and 9-antenna excitation was 42.42 °C, 42.48 °C, 42.49 °C, and 42.48 °C, respectively. The temperature induced in tumor and healthy tissues is similar for varying number of active channels. However, the amount of power consumed was 25.2%, 53.9%, 97.9% higher for 15, 12, 9 active antennas compared to the filled array. Antenna array with 12 active elements was chosen as the optimal combination as it provided selective tumor heating with good tradeoff between number of channels and power consumption. The heating ability of the thinned array was assessed for 50% reduction in the number of active antennas on patient derived heterogeneous breast phantoms for five tumor target locations. The good agreement between simulated and measured thermal distributions demonstrate the selective heating ability of our phased array applicator for 50% reduction in hardware resources. The study outcome enables us to realize a cost-effective hyperthermia treatment delivery system.

Cancers of the neck, breast, and lower extremities are common malignancies diagnosed in India with a higher incidence of advanced-stage disease. Phased array (PA) applicators reported for hyperthermia treatment (HT) of the breast have small focal region and high cross-coupling, and those reported for lower extremities provide regional heating and limited steering. In this study, we present the numerical design of site-specific PA applicators for HT of large solid tumors in the neck, breast, and lower extremities using a miniaturized 434 MHz cavity-backed water-loaded patch antenna. The fabricated antenna has 38 × 36 mm2 aperture, more than 90% power coupling, 25 MHz bandwidth, and good agreement between simulated and measured specific absorption rate (SAR) in phantom. The site-specific applicators demonstrated less power reflection (<−17.9 dB) and cross-coupling (<−26.8 dB) for 5 mm inter-ring spacing. SAR indicators for 64 cc tumor at varying locations in simplified layered three-dimensional (3D) tissue models of the neck, breast, and leg showed average power absorption ratio (aPAratio) ≥ 3.16, target to hotspot quotient (THQ) ≥ 0.57, 25% iso-SAR coverage (TC25) ≥ 81%, and 50% iso-SAR coverage (TC50) ≥51.8%. Simulation results of site-specific applicators for 3D inhomogeneous patient models showed aPAratio ≥ 5.98, THQ ≥ 0.9, TC50 ≥ 86%, and 100% TC25 for all sites. It is concluded that the 434 MHz miniaturized cavity-backed patch antenna can be used to develop high-density PA applicators with 12–24 antennas for HT of large solid tumors (≥4 cm) in the neck, breast, and lower extremities with 3D steering ability and less cross-coupling (≤−26.8 dB). © 2020 Bioelectromagnetics Society.