CT images of the whole lungs were obtained at full inspiration using 3 different scanners, including an Aquilion Prime scanner (Canon Medical Systems, Otawara, Japan) at Takeda Hospital and an Aquilion One scanner (Canon Medical Systems) and a LightSpeed VCT scanner (GE Healthcare, Waukesha, WI, USA) at Tsukuba Medical Center. In Takeda Hospital, archived images of 0.5 mm slices reconstructed with the sharp (FC53) kernel were converted to soft kernel (FC13)-based images at 1.0 mm slice thickness using a previously established deep neural network model [32 (link)]. In the Tsukuba Medical Center, images at 1.0 mm and 1.25 mm slice thicknesses reconstructed with the soft (FC02 and STANDARD) kernel and sharp (FC53 and LUNG) kernel were obtained using an Aquilion One and LightSpeed VCT scanner, respectively. The scanning conditions were 120 kVp, 0.4 s exposure time, and autoexposure control for all 3 scanners. The sharp kernel-based images were used for visual assessment of CLE and PSE, whereas the soft kernel-based images were used for quantifications of PM, SAT, ESM, EAT, and CAC.