Fig. 5
From: Image quality assessment of pediatric chest and abdomen CT by deep learning reconstruction
Chest CT images with mediastinal window of a 1-year-old girl who had a cough. a–d Axial contrast-enhanced CT images with a standard 50% adaptive statistical iterative reconstruction-V (50% ASIR-V), b 100% ASIR-V, c medium-strength deep learning image reconstruction (DLR-M), and d high-strength deep learning image reconstruction (DLR-H) show no difference in beam hardening artifacts due to dense contrast material in the superior vena cava (arrow) and motion artifacts in the bilateral ribs (arrow heads), resulting in lower reader scores for artifacts. Both readers thought the image was suboptimal. Contrast to noise ratio (CNR) of the lung was 4.8 in 50% ASIR-V, 12.0 in 100% ASIR-V, 13.0 in DLR-M, and 15.0 in DLR-H