SAR Image Reconstruction of RF Field Based on MR Temperature Image

High-resolution MRI images are essential for the accurate diagnosis of diseases. Increasing the magnetic field strength achieves this but also raises the Specific Absorption Rate (SAR) of electromagnetic waves in the body, leading to safety concerns. Currently, SAR is managed by assuming an even distribution of electromagnetic waves; however, high-field MRI can cause localized concentrations. No direct method for SAR visualization exists that relies on numerical analysis, resulting in conservative management and inefficiency. This paper applies a method for converting temperature images from magnetic resonance imaging (MRI) into SAR images using Pennes’ bioheat equation and Fourier transformation with gentle low-pass filtering. The method was tested using an agar-gel phantom heated with a heating RF coil, and the temperature changes were measured using the proton resonance frequency shift (PRFS) technique in a 3T MRI system. The results are validated by electromagnetic numerical simulations using xFDTD. Experimental results showed SAR images could be reconstructed from temperature images with a maximum error of 3.6 % compared with numerical simulations. The reconstruction took 5.2 s on a PC. This method can aid in managing electromagnetic safety in MRI and verifying the RF coil performance during development. Future studies will validate this method using different phantom configurations and in vivo experiments.