Author: Emi Ai Eastman, Christina Lee, Xinhua Li, Alexander W. Scott, Yifang (Jimmy) Zhou 👨🔬
Affiliation: Cedars-Sinai Medical Center 🌍
Purpose: Iterative reconstruction (IR) methods are valuable for reducing dose in modern CT; however, IR methods have the effect of reducing spatial resolution and hence the lesion edge sharpness. Furthermore, the resolution degradation may be dependent on the contrast, so that marginally-visible lesions may have detectability further reduced with increasing IR. We propose that for targets with contrast < 100 HU the rate of edge degradation with increasing IR strength is contrast-dependent.
Methods: A custom anthropomorphic liver phantom with two-phase iodinated lesions was scanned on a GE Revolution CT at 15mGy CTDIvol with a slice thickness of 2.5mm and a DFOV of 18cm. Images were reconstructed using ASIR-V 0-100% in 10% increments. Fourteen acquisitions were performed and averaged together using ImageJ. To determine the lesion edge sharpness, two line profiles were drawn through the center of each lesion. To achieve oversampling, the lines were at ±4° to the horizontal and points within two vertical pixels of the lines were projected onto them. Taking advantage of symmetry, each projected line was folded about the center and combined so that the lesion edge regions were aligned. To reduce the impact of noise, data points were combined based on their projected coordinate and a rolling average was performed with a 1:3:1 weighting. The data was fitted with a sigmoid (Boltzmann) curve and the derivative of the edge function (LSF) was calculated to find the FWTM. The normalized FWTM for each ASIR-V image was used to represent the rate of edge degradation at that contrast level.
Results: The normalized FWTM increased the least for highest contrast lesion (8.4%), the most for the lowest contrast lesion (24%) and intermediate contrast lesions had values between 10%-12%.
Conclusion: The edge-degradation with increasing IR strength was found to be greater for low-contrast lesions and least for high-contrast lesions.