Evaluation of the Imaging Properties of a CT Scanner with the Adaptive Statistical Iterative Reconstruction Algorithm - Noise, Contrast and Spatial Resolution Properties of CT Images Reconstructed at Different Blending Levels

X-ray Computed Tomography (CT) is an essential imaging technique for different diagnostic and therapeutic tasks. However, ionizing radiation from CT scanners represents the largest source of medical exposure for the population of industrialized countries. In order to reduce CT dose during patient examination, iterative reconstruction algorithms have been developed to help existing dose reduction methods. In this paper, we studied the image quality performance of a 64-slice CT scanner (Optima CT660, GE Healthcare, Waukesha, WI, USA) that implements both the conventional filtered back-projection (FBP) and the Adaptive Statistical Iterative Reconstruction (ASIR, GE Healthcare, Waukesha, WI, USA) algorithm. In order to compare the performance of these two reconstruction technologies, CT images of the Catphan®504 phantom were reconstructed using both conventional FBP and ASIR with different percentages of reconstruction from 20% to 100%. Noise level, noise power spectrum (NPS), contrast-tonoise ratio (CNR) and modulation transfer function (MTF) were estimated for different values of the main radiation exposure parameters (i.e. mAs, kVp, pitch and slice thickness) and contrast objects. We found that, as compared to conventional FBP, noise/CNR decreases/increases non-linearly up to 50%/100% when increasing the ASIR blending level of reconstruction. Furthermore, ASIR modifies the NPS curve shape (i.e. the noise texture). The MTF for ASIR-reconstructed images depended on both tube load and contrast level, whereas MTF of FBP-reconstructed images did not. For lower tube load and contrast level, ASIR offered lower performance as compared to conventional FBP in terms of reduced spatial resolution and MTF decreased with increasing ASIR blending level of reconstruction.