Ultimate Intrinsic SNR in the Torso of Realistic Body Models

This work aims to investigate how the ultimate intrinsic signal-to-noise ratio (uiSNR) varies with increasing static magnetic field $B_0$ in the torso of realistic body models.

A dipole cloud was positioned around the realistic body model and randomly excited. The volume integral solver MARIE was used to calculate the corresponding electromagnetic fields. The uiSNR maps were calculated using these electromagnetic bases and were fitted with the power law for different $B_0$ ranges.

The uiSNR could be reliably calculated in regions deeper than 3 cm, where convergence of uiSNR over the number of basis vectors was achieved. In a lower magnetic field range (from 0.55 to 3 T), the uiSNR increases roughly linearly versus $B_0$ with small variation throughout the torso (Ella: uiSNR ∝ B₀ ^0.96±0.07, Duke: uiSNR ∝ B₀ ^0.98±0.10). In an upper magnetic field range (from 5 to 14 T), the uiSNR increases superlinearly in the torso (Ella: uiSNR ∝ B₀ ^1.86±0.25, Duke: uiSNR ∝ B₀ ^1.99±0.28), with a larger variation correlated to the heterogeneous structure of the body model.

The superlinear scaling exponent in the upper magnetic field range indicates the promise of applying UHF MRI for body imaging.