Loading...
Triple Coaxial-Half-Slot Antenna Scheme with Deep Learning-Based Temperature Prediction for Hepatic Microwave Ablation: Finite Element Analysis and in Vitro Experiment
Author(s)
Wongketsada, Teerasak
Date Issued
January 1, 2021
Type
Article
Abstract
This research proposes a triple-antenna scheme of three coaxial half slot antennas (CHSA) for minimally invasive hepatic microwave ablation (MWA). In the MWA treatment, the antennas were positioned around the cancer tissue in equilateral triangle formation, with the half-slot side of the antennas directed at the triangle center to concentrate the microwave energy. Finite element (FE) simulations and in vitro experiments with swine liver were carried out to determine the temperatures and coagulation zone under different distances between the three antennas: 10, 20, and 30 mm. Besides, a regression-based deep learning algorithm was utilized to predict temperatures at the midpoint of the triple-CHSA arrangement without inserting a temperature sensor into the targeted tissue. The algorithm relied on temperatures at a site adjacent to the targeted cancer tissue to predict the midpoint temperatures. The triple-CHSA scheme could achieve a large coagulation zone with temperatures of 60 °C or higher. The FE simulation results showed that the largest coagulation volumes achieved were 18.425, 29.312, and 63.507 cm3 for the between-antenna distances of 10, 20, and 30 mm, respectively. The largest coagulation volumes of in vitro experiments were 18.437, 29.317, and 63.504 cm3 for the corresponding between-antenna distances. The prediction accuracy of the deep learning algorithm were 99.98 % for mean square error, 98.75 % for mean absolute error, and 99.22 % for mean absolute percentage error. The novelty of the research lies in the use of CHSA surrounding the cancer tissue in equilateral triangle formation for hepatic MWA, as opposed to the conventional hepatic MWA procedure which requires inserting antennas into the targeted tissue. Another research novelty is the use of regression-based deep learning algorithm to predict the temperature of liver tissue at the midpoint of the triple-CHSA arrangement.
Citation
IEEE Access, 9, 79572-79587, 2021