Assessment and Monitoring of RF Safety for Ultra-High Field MRI

The radio frequency (RF) energy deposited in a human subject undergoing a 7T MRI scan has the potential to cause localized tissue heating. The use of parallel transmit MRI at 7T increases the risk of localized heating due interference effects among the simultaneously transmitting channels. The degree of risk caused by RF heating is quantified by the local specific absorption rate (SAR). However, neither small temperature changes nor SAR are easy to measure in-vivo using MRI. Thus, RF safety decisions for local SAR depend on numerical simulations of electromagnetic fields derived from detailed computerized models of the RF coil (or parallel transmit array) and an anatomically accurate representation of the subject. Deviations between the simulation model and the realistic, physical MRI setup can result in drastically different local SAR predictions. Given that local SAR prediction is subject to such variability, safety is assured by assuming worst case channel interference (in the case of parallel transmit) and multiplication with a conservative safety tolerance factor. This practice often results in overly restrictive local SAR predictions, forcing MR users to either decrease the flip angle of the RF pulses in the sequence or increase the duration of the scan. The goal of this thesis was to better understand various sources of variability in local SAR prediction as well as try to improve SAR prediction in parallel transmit to reduce uncertainty. One of the studies included in this thesis assessed the variability in local SAR caused by the presence of severe anatomical abnormalities. We specifically looked at brain tumors, due to the hypothesis that increased electrical conductivity leads to increased local SAR. We observed that, while local SAR was elevated in the region of the tumor, the tumor local SAR value typically does not exceed the maximum SAR value observed elsewhere in the head. Another two studies focused on properly simulating multi-channel interference among parallel transmit array elements. Directional couplers were used to measure the amount of array element power coupling and reflections. We showed that the coupling can be included in the simulation of multi-transmit arrays to improve the accuracy of the resulting local SAR prediction. Additionally, we showed that it is possible to measure electrical current on the transmit elements using directional coupler readings, which may be advantageous for SAR predictions compared to the method of measuring the coupling. However, both methods can be used to improve local SAR assessment as well as allow for local SAR monitoring on the scanner. The last study investigates the benefit gained from using multi-transmit for prostate imaging at 3T. The 3T local transmit array dramatically reduces total power; however, the local SAR limit remains restrictive. Through this work, we gained new insights into subject-specific SAR calculation and improved the reliability of multi-transmit local SAR estimation. We hope that the knowledge gained through this thesis will lead to improved confidence in local SAR predictions, which is necessary for safe scanning and acceptance of 7T MRI/parallel transmit as a clinical diagnostic tool