Meeting Banner
Abstract #4821

A mesh-based CNN for the evaluation of MR RF-induced heating of complex-shaped passive implants

Jiajun Chang1, Jianfeng Zheng1, Ran Guo1, Qianlong Lan1, Mayur Thakore2, Wolfgang Kainz3, and Ji Chen1
1Electrical and Computer Engineering, University of Houston, Houston, TX, United States, 2Stryker Corporation, Mahwah, NJ, United States, 3High Performance Computing for MRI Safety, LLC, Jasper, GA, United States


In this research, a convolutional neural network (CNN) model is developed to predict the RF-induced heating for several tibia plate systems. One thousand four hundred and sixteen device configurations were developed, and the peak 1g-average SAR values were extracted. A subset of the data was used as training set and simulation meshes were used as the input of the CNN model. Results showed a quick network convergence and high correlation. The network also had a low absolute and percentage error level. This demonstrates that one can potentially use CNN model to predict the RF-induced heating of plate systems.

This abstract and the presentation materials are available to members only; a login is required.

Join Here