CT image quality in sinogram affirmed iterative reconstruction phantom study - is there a point of diminishing returns?

In our pediatric practice, we have observed qualitatively limited improvement in the image quality of images generated with sinogram affirmed iterative reconstruction (SAFIRE) compared to series generated with filtered back projection (FBP), particularly in cases near or below a CT dose index volume (CTDIvol) of 1-mGy. To determine whether the image quality advantage of SAFIRE remains constant across clinically used CT dose levels in an American College of Radiology (ACR) CT accreditation phantom including the lower dose range used in pediatric imaging. An exemption from institutional review board approval was obtained for this phantom-based study. An ACR quality phantom was scanned in incremental kV steps and effective tube current intervals. Acquisitions were reconstructed with FBP and SAFIRE strengths of 1, 3 and 5. Image quality measures were calculated including signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), low-contrast resolution and high-contrast resolution. Peak SNR was also calculated. Descriptive and nonparametric statistics were used to compare these image quality metrics while normalizing to CT dose index (CTDI). The percent improvement in SNR and peak SNR of SAFIRE reconstructions compared to FBP decreased from about 70% for image sets acquired above a 1.42 mGy CTDI to 25% at a 0.25 mGy CTDI. CNR improvement with SAFIRE did not vary with dose. No significant difference was seen in the low-contrast resolution or high-contrast resolution of SAFIRE images compared to FBP. SNR did not improve equally after applying SAFIRE across a spectrum clinically used CTDIs. Below a threshold CTDI, the incremental improvement of SAFIRE compared to FBP decreased.