MUMBAI, India, Nov. 2 -- Intellectual Property India has published a patent application (202621024708 A) filed by Dr. Poonam Gadge; Dr. Bhushan Suresh Deore; Dr. Vishakha Gaikwad; Nishant Wanjari; Geet Tigaiya; Shivam Singh Rathore; and Harshit Vinay Gadge on March 02, 2026, for Automated Electromagnetic Simulation And Pso-Optimized Machine Learning System For Non- Invasive Lung Tumor Prediction.
Inventors include Dr. Poonam Gadge; Dr. Bhushan Suresh Deore; Dr. Vishakha Gaikwad; Nishant Wanjari; Geet Tigaiya; Shivam Singh Rathore; and Harshit Vinay Gadge.
The application for the patent was published on October 30, 2026, under issue no. 44/2026.
Abstract: ABSTRACT The present invention discloses an automated electromagnetic simulation and machine learning framework for non-invasive lung tumor size prediction. The system integrates Finite-Difference Time-Domain (FDTD) electromagnetic modeling with Particle Swarm Optimization (PSO)-enhanced regression analysis to characterize tumor-induced dielectric variations within a multi-layer thoracic phantom. Automated scripts generate spherical tumor geometries of varying sizes and positions within lung tissue, followed by batch electromagnetic simulations under microwave excitation. Electric field distributions are computed and exported for analysis. Time-domain features including mean, maximum, root mean square electric field values, and relative specific absorption rate are extracted. Frequency-domain features such as spectral centroid, peak frequency, and band-limited power metrics are derived to capture tumor-induced spectral variations. A Particle Swarm Optimization algorithm selects an optimal subset of predictive features by minimizing cross-validated root mean squared error. The selected features are used to train Decision Tree and Random Forest regression models for tumor size estimation. The optimized ensemble model demonstrates improved predictive accuracy and nonlinear modeling capability compared to baseline approaches The invention establishes a scalable, reproducible, and physics-informed diagnostic pipeline that integrates computational electromagnetics with artificial intelligence. The system enables synthetic biomedical dataset generation and provides a foundation for future microwave-based, radiation-free tumor detection technologies suitable for early-stage lung cancer diagnostics and intelligent medical imaging systems.
Disclaimer: Curated by HT Syndication.
