Fig. 2
From: Deep learning reconstruction enhances tophus detection in a dual-energy CT phantom study
Material maps of DECT scans reconstructed with different reconstruction methods and corresponding CT images (80 kVp image data for FBP and IR, 60 keV monochromatic data for DLR). The color lookup table ‘Rainbow 2’ provided by the software (Horos v.2.2.0, The Horos Project) was used for parts a–c and g–i, with a window level of 220 and a window width of 220, corresponding to a range of 110 HU to 330 HU. Upper section (raster phantom): Concentration-color-coded 3D images of the MSU maps and corresponding CT images reconstructed with FBP (a, d), IR (b, e), and DLR (strong, c, f). The syringes, presented from left to right, contain MSU concentrations in descending order (50%, 45%, 40%, and 35%). Lower section (bio-phantom): Concentration-color-coded 3D images of the MSU maps and corresponding CT images reconstructed with FBP (g, j), IR (h, k), and DLR (strong, i, l). The syringes, containing different MSU concentrations, are arranged around the elbow joint in ascending clockwise order. An increase in true-positive detections with a decrease in false-positive artifacts is seen for DLR as compared with FBP and IR.
