Reducing electromagnetic interference in MR thermometry: A comparison of setup configurations for MR-guided microwave ablations.

Magnetic Resonance (MR) thermometry is used for the monitoring of MR-guided microwave ablations (MWA), and for the intraoperative evaluation of ablation regions. Nevertheless, the accuracy of temperature mapping may be compromised by electromagnetic interference emanating from the microwave (MW) generator. This study evaluated different setups for improving magnetic resonance imaging (MRI) during MWA with a modified MW generator. MWA was performed in 15 gel phantoms comparing three setups: The MW generator was placed outside the MR scanner room, either connected to the MW applicator using a penetration panel with a radiofrequency (RF) filter and a 7 m coaxial cable (Setup 1), or through a waveguide using a 5 m coaxial cable (Setup 2). Setup 3 employed the MW generator within the MR scan room, connected by a 5 m coaxial cable. The coaxial cables in setups 2 and 3 were modified with custom shielding to reduce interference. The setups during ablation (active setup) were compared to a reference setup without the presence of the MW system. Thermometry and thermal dose maps (CEM43 model) were compared for the three configurations. Primary endpoints for assessment were signal-to-noise ratio (SNR), temperature precision, Sørensen-Dice-Coefficient (DSC), and RF-noise spectra. Setup 3 showed highly significant electromagnetic interference during ablation with a SNR decrease by -60.4%±13.5% (p<0.001) compared to reference imaging. For setup 1 and setup 2 no significant decrease in SNR was measured with differences of -2.9%±9.8% (p=0.6) and -1.5%±12.8% (p=0.8), respectively. SNR differences were significant between active setups 1 and 3 with -51.2%±16.1% (p<0.001) and between active setups 2 and 3 with -59.0%±15.5% (p<0.001) but not significant between active setups 1 and 2 with 19.0%±13.7% (p=0.09). Furthermore, no significant differences were seen in temperature precision or DSCs between all setups, ranging from 0.33 °C ±â€¯0.04 °C (Setup 1) to 0.38 °C ±â€¯0.06 °C (Setup 3) (p=0.6) and from 87.0%±1.6% (Setup 3) to 88.1%±1.6% (Setup 2) (p=0.58), respectively. Both setups (1 and 2) with the MW generator outside the MR scanner room were beneficial to reduce electromagnetic interference during MWA. Moreover, provided that a shielded cable is utilized in setups 2 and 3, all configurations displayed negligible differences in temperature precision and DSCs, indicating that the location of the MW generator does not significantly impact the accuracy of thermometry during MWA.

Free-breathing 3D Stack-of-Stars Gradient Echo Sequence in MR-guided Percutaneous Liver Interventions: Evaluation of Workflow and Diagnostic Quality.

PURPOSE: To evaluate workflow efficiency and diagnostic quality of a free-breathing 3D stack-of-stars gradient echo (Radial GRE) sequence compared to a breath-hold 3D Cartesian gradient echo (Cartesian GRE) sequence for needle position control in MR-guided liver interventions. MATERIALS AND METHODS: 12 MR-guided liver interventions were performed on a 1.5 T Siemens Aera and analyzed retrospectively. 15 series of the Radial GRE sequence were compared to 14 series of the Cartesian GRE sequence regarding the time interval between two consecutive live-scans for needle tracking (Tracking-2-Tracking-Time). The quality of both sequences was compared by the SNR within comparable slices in liver and tumor ROIs. The CNR was calculated by subtraction of the SNR values. Subjective image quality scores of three radiologists were assessed and inter-rater reliability was tested by Fleiss' kappa. Values are given as mean ± SD. P-values < 0.05 were considered as significant. RESULTS: The median Tracking-2-Tracking-Time was significantly shorter for the Radial GRE sequence, 185 ± 42 s vs. 212 ± 142 s (p = 0.04) and the median SNR of the liver and tumor ROIs were significantly higher in the Radial GRE sequence, 249 ± 92 vs. 109 ± 67 (p = 0.03) and 165 ± 74 vs. 77 ± 43 (p = 0.02). CNR between tumor and liver ROIs showed a tendency to be higher for the Radial GRE sequence without significance, 68 ± 48 vs. 49 ± 32 (p = 0.28). Mean subjective image quality was 3.33 ± 1.08 vs. 2.62 ± 0.95 comparing Radial and Cartesian GRE with a Fleiss' kappa of 0.39 representing fair inter-rater reliability. CONCLUSION: A free-breathing 3D stack-of-stars gradient echo sequence can simplify the workflow and reduce intervention time, while providing superior image quality. Under local anesthesia, it increases patient comfort and reduces potential risks for needle dislocations in MR-guided liver interventions by avoiding respiratory arrests for needle position control.