부산대학교 도서관

This article provides an efficient and reliable alternative assessment of radio frequency (RF)-induced heating of orthopaedic implants. The concept of the heating factor is defined to model the RF response of orthopaedic implants, and an artificial neural network based on the Chebyshev parameter model-based artificial neural network (CP-ANN) is designed to achieve efficient and accurate prediction of the heating factor. The performance of the proposed CP-ANN is validated with standard simulations and measurements. The in vivo temperature of five clinical scenarios with representative orthopaedic implants in different tissue environments is evaluated using the proposed method and compared with in vitro measurements and Tier 4 simulations. The results show that with the proposed CP-ANN, the heating factors of different types of passive implants in different tissue environments can be predicted accurately (R=0.99) and efficiently (e.g., in 6 s), with more than 250-fold improvements in efficiency compared with existing algorithms. Consequently, this work can provide an efficient and reliable option for evaluating induced RF heating of orthopaedic implants under magnetic resonance imaging examination. By considering the influence of the tissue environment, this work also provides scientific insight into the potential uncertainty caused by the discrepancy between in vitro and in vivo tissue environments.