Extreme Value Theory as a Risk Analysis Tool to Infer Specific Absorption Rate Intersubject Variability Safety Margins

Specific absorption rate (SAR) in parallel transmission (pTx) ultra-high field MRI is typically evaluated by performing offline electromagnetic simulations on a generic human model. Conventionally, a safety margin that addresses the issue of intersubject variability is selected by choice of the 99th or higher percentile of the SAR distribution. In this work, we argue that this approach underestimates the risk of exceeding the SAR limit. As an alternative, we propose the use of Extreme Value Theory (EVT) for controlling the probability of extremal events in the form of large SAR deviations.