Quantification of Hepatocellular Carcinoma Vascular Dynamics With Contrast-Enhanced Ultrasound for LI-RADS Implementation

Abstract

Objective The aim of this study is to describe a comprehensive contrast-enhanced ultrasound (CEUS) imaging protocol and analysis method to implement CEUS LI-RADS (Liver Imaging Reporting and Data System) in a quantifiable manner. The methods that are validated with a prospective single-center study aim to simplify CEUS LI-RADS evaluation, remove observer bias, and potentially improve the sensitivity of CEUS LI-RADS. Materials and Methods This prospective single-center study enrolled patients with hepatocellular carcinoma (April 2021–June 2022; N = 31; mean age ± SD, 67 ± 6 years; 24 men/7 women). For each patient, at least 2 CEUS loops spanning over 5 minutes were collected for different lesion scan planes using an articulated arm to hold the transducer. Automatic respiratory gating and motion compensation algorithms removed errors due to breathing motion. The long axis of the lesion was measured in the contrast and fundamental images to capture nodule size. Parametric processing of time-intensity curve analysis on linearized data provided quantifiable information of the wash-in and washout dynamics via rise time (RT) and degree of washout (DW) parameters extracted from the time-intensity curve, respectively. A Welch t test was performed between lesion and parenchyma RT for each lesion to confirm statistically significant differences. P values for bootstrapped 95% confidence intervals of the relative degree of washout (rDW), ratio of DW between the lesion and surrounding parenchyma, were computed to quantify lesion washout. Coefficient of variation (COV) of RT, DW, and rDW was calculated for each patient between injections for both the lesion and surrounding parenchyma to gauge reproducibility of these metrics. Spearman rank correlation tests were performed among size, RT, DW, and rDW values to evaluate statistical dependence between the variables. Results The mean ± SD lesion diameter was 23 ± 8 mm. The RT for all lesions, capturing arterial phase hyperenhancement, was shorter than that of surrounding liver parenchyma (P < 0.05). All lesions also demonstrated significant (P < 0.05) but variable levels of washout at both 2-minute and 5-minute time points, quantified in rDW. The COV of RT for the lesion and surrounding parenchyma were both 11%, and the COV of DW and rDW at 2 and 5 minutes ranged from 22% to 31%. Statistically significant relationships between lesion and parenchyma RT and between lesion RT and lesion DW at the 2- and 5-minute time points were found (P < 0.05). Conclusions The imaging protocol and analysis method presented provide robust, quantitative metrics that describe the dynamic vascular patterns of LI-RADS 5 lesions classified as hepatocellular carcinomas. The RT of the bolus transit quantifies the arterial phase hyperenhancement, and the DW and rDW parameters quantify the washout from linearized CEUS intensity data. This unique methodology is able to implement the CEUS-LIRADS scheme in a quantifiable manner for the first time and remove its existing issues of currently being qualitative and suffering from subjective evaluations.