Noise reduction and spatial resolution in CT imaging with the ASIR iterative reconstruction algorithm at different doses and contrasts – a phantom study

Aims and objectives The aim of this study was to quantitatively assess noise reduction and spatial resolution in computed tomography (CT) imaging with the ASIR (Adaptive Statistical Iterative Reconstruction, GE Healthcare) reconstruction algorithm at different kVp, mAs and contrasts. Methods and materials Acquisitions of the Catphan-504 phantom were performed on a PET/CT scanner (Discovery-710, GE Healthcare). CT images were reconstructed using both filtered back projection (FBP) and ASIR with different percentages of reconstruction (20%, 40%, 60%, 80%, 100%). The image noise was estimated for different values of scanning parameters (i.e. tube-load, kilovoltage, pitch, slice thickness). Then, 3D/2D/1D noise power spectrum was estimated. Also, spatial resolution was assessed by obtaining the modulation transfer function (MTF) for a wide range of scanning parameters values and different contrast objects by the circular Edge Spread Function method (using CTP404 modulus) and the Point Spread Function method (using CTP528 modulus). . Results Image noise decreased (up to 50% as compared to FBP) with increasing the percentage of ASIR reconstruction (behaviour more relevant for higher spatial frequencies). Only for low tube load (<56 mAs) and low contrast objects (polistirene with respect to PMMA) acquisitions, MTF analysis showed that ASIR-reconstructed images were characterized by an appreciable reduction in spatial resolution, when compared to FBP-reconstructed images. Conclusion When compared to FBP, ASIR allows a relevant noise reduction without appreciably affecting image quality, except for very low dose and contrast acquisitions.