CT attenuation-based hepatic steatosis assessment using 3D organ segmentation: Impact of tube voltage and need for cutoff adjustment.
Authors
Affiliations (3)
Affiliations (3)
- Department of Radiology, Seoul National University Hospital, South Korea.
- Department of Radiology, Seoul National University Hospital, South Korea; Department of Radiology, Seoul National University College of Medicine, South Korea; Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul National University Hospital, Seoul, South Korea. Electronic address: [email protected].
- Department of Radiology, Seoul National University Hospital, South Korea; Department of Radiology, Seoul National University College of Medicine, South Korea; Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul National University Hospital, Seoul, South Korea.
Abstract
To investigate the impact of tube voltage on CT attenuation-based parameters derived from 3D organ segmentation for hepatic steatosis assessment and to evaluate the feasibility of applying tube voltage-specific cutoff values. This single-center retrospective study included 795 adults who underwent both liver CT and MR spectroscopy-proton density fat fraction (MRS-PDFF). The cohort comprised dual-energy CT (DECT; n = 398) and single-energy CT (SECT; n = 397) datasets. Using automated organ segmentation, volumetric liver attenuation and liver-minus-spleen attenuation differences were measured on pre-contrast and post-contrast images. In DECT cohort, paired 120-kVp-equivalent (120-kVp-E) and 80-kVp images were used to derive (n = 147) and validate (n = 251) linear regression equations for tube voltage adjustment, with agreement quantified using intraclass correlation coefficients (ICCs). Cutoff values for hepatic steatosis (MRS-PDFF ≥ 5 %) were derived at 120 kVp SECT dataset and applied to paired DECT data in validation dataset using tube voltage-specific adjusted cutoffs. Volumetric liver attenuation and liver-minus-spleen attenuation differences measured on both pre-contrast and post-contrast images demonstrated strong correlations between 120-kVp-E and 80-kVp images (r = 0.921-0.975; p < 0.001). In the validation dataset, regression-adjusted values showed excellent agreement with measured 80-kVp values (ICCs, 0.966-0.992), with mean differences ≤ 0.7 HU. When tube voltage-specific cutoffs were applied to paired data, corresponding sensitivity and specificity were comparable between 120-kVp-E and 80-kVp images. CT attenuation-based parameters for hepatic steatosis assessment are affected by tube voltage, and tube voltage-specific adjusted cutoff values can be applied for reliable classification across different kVp settings.