Evaluation of the Time Difference Method in Identifying Hepatocellular Carcinoma in Current CEUS LR-M Category Nodules

Li-jin Shao,Yan-jie Wang,Shan-shan Yin,Bin-bin Jiang,Yu-hui Liu,Ji-chen Wang,Wei Yang,Wei Wu,Kun Yan
DOI: https://doi.org/10.1016/j.ultrasmedbio.2023.12.010
IF: 3.694
2024-01-22
Ultrasound in Medicine & Biology
Abstract:Objective The aim of the work described here was to explore a potential method for improving the diagnostic detection of hepatocellular carcinoma (HCC) based on the contrast-enhanced ultrasound (CEUS) Liver Imaging Reporting and Data System (LI-RADS) Version 2017. Methods We retrospectively evaluated 585 liver nodules in 427 patients at risk for HCC from December 2020 to March 2023. The nodules were categorized as LR-1 to LR-M based on CEUS LI-RADS Version 2017 and were randomly subclassified into a developmental cohort (DC) and a validation cohort (VC) at 3:1. In the DC, the cutoff value of the time difference (∆ T ) for differentiating HCC from other malignancies by LR-M was calculated and used to reclassify nodules in the VC. The diagnostic effect on HCC detection before and after reclassification was further assessed. Results According to the current CEUS LI-RADS, 140 of 426 (32.9%) confirmed HCC nodules were misclassified as LR-M. In the DC (439 nodules), the receiver operating characteristic (ROC) curve revealed that the cutoff value of ∆ T (wash-out onset time minus contrast arrival time) recommended for HCC diagnosis was greater than 21 s. In the VC (146 nodules), 34 HCCs were correctly categorized as LR-5 according to the cutoff value, and after reclassification, LR-5 had higher accuracy (67.1% vs. 89.0%, p < 0.001) and sensitivity (56.0% vs. 87.2%, p < 0.001) for HCC diagnosis with high specificity (100% vs. 94.6%, p = 0.500). Conclusion Using the time difference method could identify HCC nodules misdiagnosed as LR-M and improve the diagnostic performance of current CEUS LI-RADS.
radiology, nuclear medicine & medical imaging,acoustics
What problem does this paper attempt to address?