[en] To determine the value of lesion hypointensity in the hepatobiliary phase (HBP) on gadobenate dimeglumine-enhanced MRI as an additional major imaging feature for diagnosis of hepatocellular carcinoma (HCC) using LI-RADS v2018 criteria. Between March 2016 and August 2018, 235 patients with 250 hepatic nodules at high risk of HCC underwent gadobenate dimeglumine-enhanced MRI. Two radiologists independently evaluated the imaging features and classified the nodules based on LI-RADS v2018 criteria, and their consensus data were used to calculate the diagnostic performance of LI-RADS categories. Two modified LI-RADS definitions were as follows: (1) LI-RADS-m1: HBP hypointensity as an additional major feature; (2) LI-RADS-m2: HBP hypointensity as an alternative to ''enhancing capsule'' as an additional major feature. The diagnostic performance of LR-5 categories was compared using McNemar's test. The sensitivity and specificity for LR-5 classification using original LI-RADS v2018 criteria were 78.1% and 96.3%, respectively. Significantly improved sensitivity (82.7%; p = 0.004) with unchanged specificity (96.3%; p = 1.00) was seen for LR-5 classification using LI-RADS-m1. Similar sensitivity and specificity (82.7% and 96.3%, respectively) were also seen using LI-RADS-m2. Significantly improved sensitivity (79.5% vs. 64.0%; p = 0.031) with unchanged specificity (96.2% vs. 96.2%, p = 1.00) was seen using both LI-RADS-m1 and LI-RADS-m2 compared to the original LI-RADS v2018 for 39 HCC nodules measuring 10-19 mm. Lesion hypointensity on gadobenate dimeglumine-enhanced HBP MRI may improve sensitivity for LR-5 classification beyond that achievable using conventional LI-RADS v2018 criteria. Lesion hypointensity may prove a suitable alternative imaging feature to enhancing capsule for accurate LR-5 classification.

[en] 

Purpose

To evaluate whether a three-phase dynamic contrast-enhanced CT protocol, when combined with a deep learning model, has similar accuracy in differentiating hepatocellular carcinoma (HCC) from other focal liver lesions (FLLs) compared with a four-phase protocol.

Methods

Three hundred and forty-two patients (mean age 49.1 ± 10.5 years, range 19–86 years, 65.8% male) scanned with a four-phase CT protocol (precontrast, arterial, portal-venous and delayed phases) were retrospectively enrolled. A total of 449 FLLs were categorized into HCC and non-HCC groups based on the best available reference standard. Three convolutional dense networks (CDNs) with the input of four-phase CT images (model A), three-phase images without portal-venous phase (model B) and three-phase images without precontrast phase (model C) were trained on 80% of lesions and evaluated in the other 20% by receiver operating characteristics (ROC) and confusion matrix analysis. The DeLong test was performed to compare the areas under the ROC curves (AUCs) of A with B, B with C, and A with C.

Results

The diagnostic accuracy in differentiating HCC from other FLLs on test sets was 83.3% for model A, 81.1% for model B and 85.6% for model C, and the AUCs were 0.925, 0.862 and 0.920, respectively. The AUCs of models A and C did not differ significantly (p = 0.765), but the AUCs of models A and B (p = 0.038) and of models B and C (p = 0.028) did.

Conclusions

When combined with a CDN, a three-phase CT protocol without precontrast showed similar diagnostic accuracy as a four-phase protocol in differentiating HCC from other FLLs, suggesting that the multiphase CT protocol for HCC diagnosis might be optimized by removing the precontrast phase to reduce radiation dose.

[en] To evaluate the value of gadobenate dimeglumine-enhanced MRI in predicting the pathologic grade of hepatocellular carcinoma (HCC). Patients with pathologically proven HCC who underwent preoperative gadobenate dimeglumine-enhanced dynamic MRI were included. Two radiologists blinded to pathology results evaluated images in consensus. Lesions were evaluated quantitatively in terms of ratio of enhancement (RE), and qualitatively based on image features related to tumor aggressiveness. Logistic regression and ROC analyses were used to determine the value of these parameters to predict pathologic grade. In total, 221 patients (194 males, 27 females, aged 52.9$\pm <!--\; ??\; -->$11.7 years) with 49 poorly differentiated HCCs and 172 well/moderately differentiated HCCs were evaluated. Features significantly related to poorer pathologic grade at univariate analysis included lower RE in the early arterial phase (EAP) (p = 0.001), nonsmooth margins (p = 0.001), absence of capsule (p < 0.001), arterial peritumoral hyperenhancement (p < 0.001), higher AFP (p = 0.004), multiple tumors (p = 0.026), and larger tumor size (p = 0.028). At multivariate analysis, lower RE (EAP) (OR = 0.144, p = 0.002), absence of capsule (OR = 0.281, p = 0.004), and arterial peritumoral hyperenhancement (OR = 4.117, p < 0.001) were independent predictive factors for poorer pathologic grade. ROC analysis showed lower RE (EAP) was predictive of poorer pathologic grade (AUC = 0.667). AUC increased to 0.797 when combined with absence of capsule and presence of peritumoral hyperenhancement. Lower RE (EAP), absence of capsule, and arterial peritumoral hyperenhancement were predictive biomarkers for poorer pathologic grade of HCC on gadobenate dimeglumine-enhanced dynamic MRI.