Author: Chao Guo, Xinhua Li, Michael F. McNitt-Gray, Di Zhang, Yifang (Jimmy) Zhou 👨🔬
Affiliation: UCLA, David Geffen School of Medicine at UCLA, Cedars-Sinai Medical Center 🌍
Purpose: The objective was to quantify the liver fat volume fraction (FVF) with virtual monochromatic imaging (VMI) using photon-counting CT (PCCT) and dual energy CT (DECT).
Methods: A custom designed liver phantom (size: 25 x 32.5 cm) was fabricated. The phantom contains six ellipsoidal fatty lesions of known FVF (5- 40%) in the liver. The phantom was scanned at full collimation (5.76 cm) with typical clinical doses, CTDIvol of 14 mGy and 8.5 mGy, on a Siemens Force DECT and a Siemens Naeotom Alpha PCCT, respectively. 80 kV/150 kV(Sn) and 100 kV/150 kV(Sn) were chosen on the DECT, and 120 kV was used with the quantum plus mode on the PCCT. Virtual monochromatic images (3 mm width) were reconstructed from 40 to 140 keV. Volumes of interest were used to measure the Hounsfield units (HU) of the lesions. The final HU from the repeats were correlated with the FVF.
Results: For the DECT of 80/150 kV(Sn), the theoretically predicted linear model for FVF-HU resulted in the least FVF root-mean-square (R.M.S.) errors of 2.38- 3.08% from 100-140 keV. For the PCCT at 120 kV, the smallest R.M.S. errors were 5.85- 6.13% at lower VMI energies from 80 – 100 keV. For comparison, the best R.M.S. errors from the DECT 100/150 kV(Sn) and from the single energy at 120 kV were 10.27%, and 9.55%, respectively.
Conclusion: The photon counting CT resulted in appreciably less accurate FVF than from the dual energy CT at the widest spectrum separation. As the cut-off FVF value for non-alcoholic liver disease is only slightly above 5%, this finding is clinically important.