Electromagnetic Wave Absorption in the Human Head: A Virtual Sensor Based on a Deep-Learning Model

Area of Science:

• Biomedical Engineering
• Computational Electromagnetics
• Artificial Intelligence

Background:

• Accurate assessment of electromagnetic wave energy absorption in the human body is crucial for wireless system safety analysis.
• Traditional numerical methods (e.g., finite-difference time-domain) based on Maxwell's equations are computationally intensive, particularly at high frequencies requiring fine mesh discretization.

Purpose of the Study:

• To develop a computationally efficient surrogate model for predicting electromagnetic wave absorption in the human body.
• To utilize deep learning, specifically a Convolutional Neural Network (CNN), to estimate power density in human head models.

Main Methods:

• A Convolutional Neural Network (CNN) was trained using data generated from finite-difference time-domain (FDTD) simulations.
• The CNN was designed to predict average and maximum power density in the human head cross-section at 3.5 GHz.
• The model was applied to estimate absorption in the entire head and specific regions like the eyeballs.

Main Results:

• The developed deep learning surrogate model provides rapid estimation of electromagnetic wave power density.
• The method accurately determines average and maximum power density for the head and eyeball regions.
• Results from the CNN model closely align with those obtained from traditional Maxwell's equations-based methods.

Conclusions:

• Deep learning offers a significantly faster alternative to conventional numerical methods for analyzing electromagnetic wave absorption in human tissues.
• The proposed CNN-based surrogate model is effective for quick and accurate assessment of radiofrequency energy deposition in the human head.
• This approach enhances the efficiency of safety assessments in the design and analysis of wireless communication systems.