Hepatic Steatosis Assessment with Novel Ultra-Low Dose Chest CT

Objective: Our study evaluated reliability of fatty liver assesment using a novel ultra-low dose chest CT (ULDCT) protocol.

Materials and Methods: Institutional review board approved our prospective study of patients who underwent normal dose chest CT (NDCT) and ULDCT. ULDCT was performed with average radiation dose < 5% of NDCT, both without contrast media. Artificial intelligence (AI) based denoising methods produced two denoised ULDCT (dULDCT) series. Hepatic and splenic attenuation was measured in NDCT, ULDCT and dULDCT. Considering NDCT the comparator scan, regression analyses assessed performance of ULDCT and dULDCT for fatty liver criteria: liver attenuation 10HU than liver attenuation.

Results: Six (8.3%) out of 72 included patients were considered to have fatty liver on NDCT according to all 3 criteria. The ULDCT series had specificity over 98.5% in fatty liver detection, with sensitivity over 50% and accuracy over 94.3%. Both denoising series had 100% sensitivity in detecting fatty liver when considering liver attenuation<40HU, with sensitivities >77.3% and accuracies >79.1%. The Wilcoxon test, compared liver attenuation measurements between the comparator scan and the ULDCT series, yielded significant differences (p<=0.047).

Significance: Although attenuation measurements were significantly different between NDCT and ULDCT, the specificity and accuracy for detecting fatty liver were high. Hepatic steatosis could be evaluated on ULDCT using each of the three acceptable criteria on ULDCT with high specificity and accuracy, but may miss some patients with fatty liver. The dULDCT showed capability of fatty liver detection with high sensitivity when using the classic rule of liver attenuation <40HU, which may overcall fatty liver in some patients. Further investigation with a larger study cohort is needed to validate these findings, especially given the low prevalence of patients with fatty liver in our study