Machine Learning Approach in Optimal Localization of Tumor Using a Novel SIW-Based Antenna for Improvement of Ablation Zone in Hepatocellular Carcinoma

Abstract

In recent years, hepatocellular carcinoma has been the leading cause of cancer-related mortality and thermal ablation procedures such as MWA (microwave ablation) and RFA (Radiofrequency ablation) provide a viable alternative to radiation and surgical excision. Due to uneven ablation and tissue charring in RFA, MWA offers faster and uniform heating as a result of its higher operating frequency. In MWA, the antenna probe heats the tissue to the point of cell necrosis using electromagnetic heating. In this aspect, a substrate-integrated waveguide (SIW) antenna is designed to work at 2.45 GHz. Further, a finite element method (FEM) is employed to model the probe and the liver tissue environment. Within 10 minutes of application time, a maximum ablation diameter of 32 mm (transversal-T) and 25 mm (axial-A) is achieved at 20W of power. Since the position of the applicator probe is of utmost importance, the relationship between the percentage of ablation, probe tip position, and tumor diameter is evaluated using ML- algorithm to estimate the ideal probe location for maximal tumor ablation and thus may improve clinical outcome.

Keywords

• Ablation zone
• cat boost
• microwave ablation antenna
• support vector regression
• XG boost