SU-E-I-68: The Effect of Partially Parallel Imaging on SNR Across Scanning Platforms.

Abstract

Different manufacturers implement partially parallel imaging (PPI) with different proprietary reconstruction methods and algorithms. Additionally, signal to noise ratio (SNR) associated with PPI is highly dependent on coil design and conformation, as quantified by the g-factor. This potentially leads to differences in SNR across coils and scanners. This study seeks to characterize the impact of PPI on SNR across four MRI platforms. Turbo spin echo (TSE) scans of a spherical phantom filled with an aqueous solution were acquired in a head coil on a 3T Siemens Tim Trio, 3T Philips Achieva, 1.5T Toshiba Vantage and 3T GE Signa Excite HD. Scans at each available integer R-value and reconstruction method were taken on each scanner. SNR was measured in each image using the difference method. SNR was normalized to a maximum value of 1.0 for each reconstruction method. Linear regressions and percent change in SNR relative to R-value were determined for each reconstruction method. Coefficient of variation (CV) was used to determine the relationship between PPI implementations. Slopes of the linear regressions ranged from ×0.24 to ×0.27. Percent SNR on average was 67.3%±3.7%, 43.3%±2.9% and 27.4%±6.9% of the associated non-PPI scan at R=2, 3 and 4 respectively. The range of percent SNR relative to non-PPI scans was 62.1%-71.7%, 39.8%-46.8% and 20.7%-34.5% at R=2, 3 and 4 respectively. The CV was 7.75% at R=2, 16.05% at R=3 and 21.79% at R=4. Slopes of the linear regressions ranged from -0.24 to -0.27. Percent SNR on average was 67.3%±3.7%, 43.3%±2.9% and 27.4%±6.9% of the associated non-PPI scan at R=2, 3 and 4 respectively. The range of percent SNR relative to non-PPI scans was 62.1%-71.7%, 39.8%-46.8% and 20.7%-34.5% at R=2, 3 and 4 respectively. The CV was 7.75% at R=2, 16.05% at R=3 and 21.79% at R=4.