Experimental and numerical optimization modelling to reduce radiofrequency-induced risks of magnetic resonance examinations on leaded implants

Abstract

Convex formulations can be used to reduce the local specific absorption rate enhancement by active medical implants of radiofrequency fields in magnetic resonance examinations while minimizing the loss of image quality. This paper demonstrates that such an optimization methodology, previously presented for strictly computational models, can be extended to a hybrid scheme using experimentally determined implant models and pre-computed fields, which can enable quasi real-time exposure optimization. The methodology determines the optimum radiofrequency field shimming condition by considering both the reduction of specific absorption rate enhancement at the tip of the implant lead, created by the interaction of the radiofrequency fields tangential to the implant trajectory with the characteristic response of the implant, and the preservation of magnetic field homogeneity, which correlates to image quality. The inputs to this workflow are those required for each implant by standard ISO 10974 evaluation, namely the validated piece-wise transfer function of the implant, the clinical routing within the patient, and the pre-computed numerical estimation of patient exposure without the implant. Optimized incident field conditions were computed to meet a range of numerical targets for specific absorption rate reduction, stepping down percentagewise from the maximum field homogeneity to the minimum exposure enhancement, for a generic implant with a flexible wire in a standard benchtop radiofrequency coil and phantom. Measurements of the corresponding specific absorption rate enhancements validated the predictions from the optimization approach within the combined confidence interval.