Digital Twin Modeling of MR Gradient Coils for Accurate Gradient Safety Evaluation of Implantable Medical Devices.

Abstract

Even with identical coil sizes and gradient strengths, differences in coil design can cause variations in field distribution, affecting gradient safety assessment of magnetic resonance imaging (MRI) for implanted devices. This study proposed a digital twin approach for gradient coil design to enable refined evaluation of gradient-induced risks. Based on the target field method, additional specific field constraints (SFC) were applied to all magnetic field components outside the region of linearity (ROL). Magnetic field distributions of three clinical axial coils were measured using a custom probe and a tailored field acquisition strategy. The measured data were partially incorporated as design constraints, with some data used to validate the model. The resulting digital twin model was applied to predict gradient-induced voltages (GIVs) along a simulated deep brain stimulation (DBS) lead path, with experimental measurements used for comparison. The results indicated that incorporating SFC reduced the discrepancy between the simulated and measured fields by over 84.5% compared with conventional designs. The linear regression R2 between predicted and measured GIVs for the X-, Y- and Z-axis gradient coils was 0.985, 0.939, and 0.973, respectively, with maximum prediction errors below 0.18 V. The proposed MR gradient digital twin method provides a helpful framework for constructing clinically relevant gradient testing environments and supports future studies on refined gradient-related safety evaluation.