Quantification of hepatic steatosis with dual-energy computed tomography: comparison with tissue reference standards...

The aim of this study was to compare dual-energy computed tomography (DECT) and magnetic resonance imaging (MRI) for fat quantification using tissue triglyceride concentration and histology as references in an animal model of hepatic steatosis.

This animal study was approved by our institution's Research Animal Resource Center. After validation of DECT and MRI using a phantom consisting of different triglyceride concentrations, a leptin-deficient obese mouse model (ob/ob) was used for this study. Twenty mice were divided into 3 groups based on expected levels of hepatic steatosis: low (n = 6), medium (n = 7), and high (n = 7) fat. After MRI at 3 T, a DECT scan was immediately performed. The caudate lobe of the liver was harvested and analyzed for triglyceride concentration using a colorimetric assay. The left lateral lobe was also extracted for histology. Magnetic resonance imaging fat-fraction (FF) and DECT measurements (attenuation, fat density, and effective atomic number) were compared with triglycerides and histology.

Phantom results demonstrated excellent correlation between triglyceride content and each of the MRI and DECT measurements (r(2) ≥ 0.96, P ≤ 0.003). In vivo, however, excellent triglyceride correlation was observed only with attenuation (r(2) = 0.89, P < 0.001) and MRI-FF (r(2) = 0.92, P < 0.001). Strong correlation existed between attenuation and MRI-FF (r(2) = 0.86, P < 0.001). Nonlinear correlation with histology was also excellent for attenuation and MRI-FF.

Dual-energy computed tomography (CT) data generated by the current Gemstone Spectral Imaging analysis tool do not improve the accuracy of fat quantification in the liver beyond what CT attenuation can already provide. Furthermore, MRI may provide an excellent reference standard for liver fat quantification when validating new CT or DECT methods in human subjects.