Each year approximately 300,000 patients with medical implants including deep brain stimulation (DBS) devices are denied magnetic resonance imaging (MRI) examination due to safety concerns ¹ . One of the major contraindications of MRI for DBS imaging is due to the potential for permanent injuries from excessive tissue heating. Consequently, post-operative MRI of DBS patients is currently approved by the FDA with specific and limited conditions of use (e.g., 1.5T only) ². One open question when evaluating RF-induced heating with DBS is the effect of the lead path and the need for patient-specific information. We report here, preliminary results of SAR maps using patient-specific lead paths extracted from post-operative CT images, and compare them to simplified lead trajectories.Finite Element Time Domain technique (FDTD) and Finite Element Method (FEM) were used to calculate local SAR values around the DBS implants at 3T MRI with a transmit head coil in two models based on patient-specific data (see Fig.1). Realistic DBS lead trajectories were extracted from post-operative CT images, and were used to construct detailed numerical models of DBS lead trajectories, and to calculate 1g- and 10g-averaged SAR. Electrically homogeneous and electrically heterogeneous head models were used for the study. Case A: FDTD simulations were performed on a 1x1x1 mm MRI-based head model. The acquisition, processing and segmentation of the head model are described in a previous work ³. Two bilateral implants were modeled. Implants were manually segmented from the CT image and adjusted onto the multi-resolution head model by applying a transformation obtained from non-linear registration (CT image to head model). Each implant was modeled by a polyline containing 500 wire segments. The first segment of wire started from subthalamic nucleus (STN) and continued up to the skull. The modeled wire was insulated except for the 2 mm in the first segment contacting the STN, to model the electrical contact. Case B: FEM simulations were performed on an electrically homogeneous head model and a uni-lateral DBS implant, built from post-operative CT images of another patient. The lead trajectory was manually segmented from CT images and a detailed model of DBS lead composed of four electrode contacts, connected through a solid core and surrounded by hollow insulation, was constructed similar to those described in previous studies ⁴. For both cases, we also constructed simplified DBS lead models in which loops where eliminated from lead trajectories. Simulation results were normalized to produce a total head averaged SAR of 3.2 W/kg in the head model without the implant. For both studies, B₁⁺ fields were calculated as B₁⁺=0.5(B_1x+jB_1y) and SAR ratios between the simplified models and realistic models were defined as SAR_Ratio=10log10(Simplified SAR/Realistic SAR).B₁⁺ fields, averaged over whole head without the implant, were comparable in both studies (3.5 µT in Case A and 3.0 µT in Case B). In both cases, we found that the simplified path, without loops, led to a much higher local SAR near the lead. Specifically, the simplified lead model in Case B showed an increase of 1g-avg.SAR and 10g-avg. SAR of 84% compared to the realistic path. The simplified lead model in Case A overestimated the peak of 1g-avg. SAR and 10g-avg. SAR by approximately 60%. To the authors’ knowledge, this work is the first attempt to evaluate quantitative values of local SAR in the brain tissue based on patient-derived realistic models of DBS lead geometries. Our results have two significant implications: Firstly, if verified in larger patient cohorts, similar studies might justify the use of simplified lead trajectories in numerical simulations that aim to assess safety of MR Conditional devices. Second, we found a significant difference between SAR values predicted from simplified models and those predicted from realistic DBS paths. Our preliminary results suggest that actual SAR values during MRI of DBS patients might be less than those estimated by simple models. However, additional validation of these results is needed. We are now in the process of evaluating a cohort of 20 DBS patients to complement this study.