Purpose: To evaluate diagnostic performance of contrast-enhanced (CE) ultrasonography (US) and microflow (MF) imaging in differentiation of atypical hepatocellular carcinoma (HCC) from focal nodular hyperplasia (FNH).
Materials and methods: Institutional review board approval was obtained, and informed consent was waived. A total of 103 patients (mean age, 43.9 years; age range, 17-75 years) were included; 38 patients with HCC and 65 with FNH underwent CE US, and vascular architecture pattern (VAP) and arrival time parametric (ATP) images were analyzed. Resident and staff radiologists independently and retrospectively reviewed CE US, VAP, and ATP images. χ(2) test and logistic regression analysis were applied to identify specific features of FNH or HCC on CE US and MF images. To compare diagnostic performance of CE US with or without MF imaging, four sets of criteria were assigned: (a) routine CE US alone, (b) VAP and CE US, (c) ATP imaging and CE US, and (d) all three methods in combination. Sensitivity, specificity, accuracy, and area under the receiver operating characteristic curve (AUC) of resident and staff radiologists were analyzed. Cohen κ statistic was used to assess agreement of CE US and MF imaging features between staff and resident radiologists.
Results: MF imaging offered significant improvements over most detection rates achieved with routine CE US signs in both groups. For staff radiologists, AUCs from criteria sets 3 (AUC = 0.873, P < .05) and 4 (AUC = 0.887, P < .05) were significantly higher than AUC from criteria set 1 (AUC = 0.835). For resident radiologists, specificity (71% and 69% vs 25%, P < .01) and accuracy (78% and 79% vs 50%, P < .01) of criteria sets 3 and 4 were significantly higher than those of criteria set 1. Moreover, AUCs for criteria sets 2 (AUC = 0.728, P < .05), 3 (AUC = 0.823, P < .01), and 4 (AUC = 0.857, P < .01) were significantly higher than those for criteria set 1 (AUC = 0.667).
Conclusion: When compared with routine CE US, MF imaging can more effectively depict specific features and offers improved diagnostic performance in the differentiation of atypical HCC from FNH, especially when used by resident radiologists. Online supplemental material is available for this article.
RSNA, 2015