• dual-energy CT
• hepatocellular carcinoma
• spectral imaging

Purpose: To retrospectively evaluate image quality and diagnostic accuracy of dual-energy CT (DECT) in the diagnosis of hepatocellular carcinoma (HCC) in a series of transplanted cirrhotic patients.
Materials and Methods: Institutional Review Board approval was obtained. The study included all consecutive cirrhotic patients who underwent liver transplantation (LTx) <120 days after multiphasic DECT with spectral imaging in late arterial phase from March 2013 to December 2014. Three sets of images (A: 140kVp polychromatic; B: 70keV monochromatic; C: “iodine-based” material decomposition images) were reviewed, assessing image quality (scale 1-5), lesion conspicuity (scale 1-3) and lesion-to-liver contrast-to-noise ratio (CNR). Using the pathological findings on the explanted livers as gold standard, sensitivity, specificity, positive and negative predictive values and accuracy were assessed and compared by DeLong method.
Results: Fifty-three patients (41 males; mean age 54.4 ± 8.2 years) were included, with 31 HCC nodules identified at pathology in 23 (43.4%) patients (mean diameter 19.1 ± 8.5 mm). Group B showed significantly higher image quality (4.91 ± 0.35) compared to groups A (4.83 ± 0.54; P=.04) and C (4.70 ± 0.57; P=.0035). Lesion conspicuity scores were 2.58 ± 0.65, 2.83 ± 0.48 and 2.92 ± 0.28 for groups A, B and C, respectively; the difference was significant comparing group A and C (P=.03). CNR was significantly higher in group C (4.72 ± 2.94) compared to groups A (2.47 ± 1.5; P<.0001) and B (3.44 ± 1.63; P=.02). On the nodule-by-nodule and patient-by-patient analyses, group C had the highest diagnostic accuracy (area-under-the-curve, AUC 0.90 and 0.96, respectively), compared to groups A (AUC 0.81 and 0.87) and B (AUC 0.87 and 0.92); the difference was significant comparing groups A and C (P=.009 and .039).
Conclusions: DECT with spectral imaging provides high quality images; by increasing CNR, iodine-based images are able to increase sensitivity in HCC diagnosis with >90% diagnostic accuracy.