Statistical simulation of SAR variability with geometric and tissue property changes by using the unscented transform

Area of Science:

• Medical Imaging Physics
• Electromagnetic Field Theory
• Computational Electromagnetics

Background:

• Local specific absorption rate (SAR) is a critical safety parameter for radio frequency (RF) transmit coils used in Magnetic Resonance Imaging (MRI).
• Variability in tissue properties and coil geometry can significantly influence local SAR, necessitating robust assessment methods.
• Accurate estimation of local SAR is essential for ensuring patient safety during MRI procedures.

Purpose of the Study:

• To introduce and validate a statistical simulation approach for evaluating local SAR variability in RF transmit coils.
• To address the impact of tissue property and geometric variations on local SAR predictions.
• To provide a method for determining local SAR limits based on statistical analysis.

Main Methods:

• Local SAR was modeled as a nonlinear transformation with random input variables representing tissue properties and geometry.
• The unscented transform was employed for efficient statistical analysis, utilizing a small set of deterministic sample points.
• Finite-difference time-domain (FDTD) electromagnetic simulations, accelerated by graphic processing units (GPUs), were used for analysis.

Main Results:

• The method was applied to a 7 Tesla (T) square loop coil for spine imaging and a 16-element brain imaging array.
• Local SAR variability was successfully examined concerning variations in tissue properties and geometric parameters.
• Local SAR limits were established by considering the mean and standard deviation of the simulated SAR values.

Conclusions:

• The proposed statistical simulation approach is efficient and versatile for studying local SAR variability.
• This method aids in the accurate assessment of RF coil safety by accounting for biological and physical variations.
• The findings support the development of safer and more reliable MRI technologies.