Using U-Nets to Predict the Effects of Head Motion on Simulated Specific Absorption Rate for Ultra-High Field...

Area of Science:

• Medical Imaging
• Magnetic Resonance Imaging
• Computational Electromagnetics

Background:

• Ultrahigh-field MRI (UHF-MRI) necessitates stringent management of specific absorption rate (SAR) due to its dependence on subject position.
• Subject motion during scanning can significantly increase local SAR, requiring large safety margins that compromise imaging performance.

Purpose of the Study:

• To develop a U-Net architecture for real-time adaptation of SAR safety calculations to subject motion.
• To enable high-performance UHF-MRI scanning without compromising patient safety, particularly for subjects unable to remain still.

Main Methods:

• Electromagnetic simulations were conducted using five body models and an 8-channel parallel-transmit coil.
• U-Nets were trained to estimate motion-induced local SAR effects by mapping Q-matrices to SAR distributions and back.
• The approach was validated by comparing network-estimated SAR with ground truth for realistic parallel-transmit pulses.

Main Results:

• Subject motion significantly impacts local SAR, but the U-Net models accurately predicted post-motion SAR distributions.
• The proposed method reduced the required safety margin from 2.14-fold to 1.3-fold.
• Imaging performance was improved to 68% of maximum, compared to 21% with static safety models accounting for all positions.

Conclusions:

• The position-aware SAR calculation method allows for reduced safety margins in UHF-MRI.
• This approach has the potential to significantly enhance scanning performance for mobile subjects while maintaining safety.