Diagnostic performance of CT/MRI LI-RADS v2018 in non-cirrhotic hepatitis C virus infection.

PURPOSE: To evaluate the diagnostic performance of LI-RADS among patients with non-cirrhotic hepatitis C virus (HCV) infection. METHODS: This retrospective, IRB-approved, single-center study included 66 observations from 43 adult patients (11 women, 32 men; median age 65 years). All patients received liver protocol CT or MRI from 2010 to 2023, had HCV, and did not have cirrhosis based on histopathology. Three board-certified abdominal radiologists blinded to histopathology and imaging follow-up assessed each observation for major features and final LI-RADS category, and inter-reader agreements with weighted kappa were calculated. The positive predictive value, sensitivity, specificity, and accuracy of in diagnosing HCC and overall malignancy was calculated. RESULTS: Of the 66 observations, 53 (80%) were malignant and 13 (20%) were benign. Positive predictive value for HCC was 0-0% for LR-1, 0-0% for LR-2, 0-33% for LR-3, 57-100% for LR-4, 98-100% for LR-5, 25-50% for LR-M, and 83-100% for LR-TIV. Positive predictive value for overall malignancy was 0-0% for LR-1, 0-0% for LR-2, 0-33% for LR-3, 57-100% for LR-4, 98-100% for LR-5, 100-100% for LR-M, and 100-100% for LR-TIV. For LR-5 in identifying HCC, sensitivity ranged from 74 to 90%, specificity from 94 to 100%, and accuracy from 80 to 91%. For the composite of LR-5, LR-M, or LR-TIV in identifying overall malignancy, sensitivity was 87-98%, specificity was 92-100%, and accuracy was 89-97%. The inter-reader agreement for major features varied from moderate to substantial, with substantial agreement for the final category. CONCLUSION: CT/MRI LI-RADS v2018 criteria can be applied to non-cirrhotic HCV patients with near-perfect specificity.

A non-invasive predictive model based on multimodality ultrasonography images to differentiate malignant from benign focal liver lesions.

We have developed a non-invasive predictive nomogram model that combines image features from Sonazoid contrast-enhanced ultrasound (SCEUS) and Sound touch elastography (STE) with clinical features for accurate differentiation of malignant from benign focal liver lesions (FLLs). This study ultimately encompassed 262 patients with FLLs from the First Hospital of Shanxi Medical University, covering the period from March 2020 to April 2023, and divided them into training set (n = 183) and test set (n = 79). Logistic regression analysis was used to identify independent indicators and develop a predictive model based on image features from SCEUS, STE, and clinical features. The area under the receiver operating characteristic (AUC) curve was determined to estimate the diagnostic performance of the nomogram with CEUS LI-RADS, and STE values. The C-index, calibration curve, and decision curve analysis (DCA) were further used for validation. Multivariate and LASSO logistic regression analyses identified that age, ALT, arterial phase hyperenhancement (APHE), enhancement level in the Kupffer phase, and Emean by STE were valuable predictors to distinguish malignant from benign lesions. The nomogram achieved AUCs of 0.988 and 0.978 in the training and test sets, respectively, outperforming the CEUS LI-RADS (0.754 and 0.824) and STE (0.909 and 0.923) alone. The C-index and calibration curve demonstrated that the nomogram offers high diagnostic accuracy with predicted values consistent with actual values. DCA indicated that the nomogram could increase the net benefit for patients. The predictive nomogram innovatively combining SCEUS, STE, and clinical features can effectively improve the diagnostic performance for focal liver lesions, which may help with individualized diagnosis and treatment in clinical practice.

Is concurrent LR-5 associated with a higher rate of hepatocellular carcinoma in LR-3 or LR-4 observations? An individual participant data meta-analysis.

BACKGROUND: The Liver Imaging Reporting and Data System (LI-RADS) does not consider factors extrinsic to the observation of interest, such as concurrent LR-5 observations. PURPOSE: To evaluate whether the presence of a concurrent LR-5 observation is associated with a difference in the probability that LR-3 or LR-4 observations represent hepatocellular carcinoma (HCC) through an individual participant data (IPD) meta-analysis. METHODS: Multiple databases were searched from 1/2014 to 2/2023 for studies evaluating the diagnostic accuracy of CT/MRI for HCC using LI-RADS v2014/2017/2018. The search strategy, study selection, and data collection process can be found at https://osf.io/rpg8x . Using a generalized linear mixed model (GLMM), IPD were pooled across studies and modeled simultaneously with a one-stage meta-analysis approach to estimate positive predictive value (PPV) of LR-3 and LR-4 observations without and with concurrent LR-5 for the diagnosis of HCC. Risk of bias was assessed using a composite reference standard and Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2). RESULTS: Twenty-nine studies comprising 2591 observations in 1456 patients (mean age 59 years, 1083 [74%] male) were included. 587/1960 (29.9%) LR-3 observations in 1009 patients had concurrent LR-5. The PPV for LR-3 observations with concurrent LR-5 was not significantly different from the PPV without LR-5 (45.4% vs 37.1%, p = 0.63). 264/631 (41.8%) LR-4 observations in 447 patients had concurrent LR-5. The PPV for LR-4 observations with concurrent LR-5 was not significantly different from LR-4 observations without concurrent LR-5 (88.6% vs 69.5%, p = 0.08). A sensitivity analysis for low-risk of bias studies (n = 9) did not differ from the primary analysis. CONCLUSION: The presence of concurrent LR-5 was not significantly associated with differences in PPV for HCC in LR-3 or LR-4 observations, supporting the current LI-RADS paradigm, wherein the presence of synchronous LR-5 may not alter the categorization of LR-3 and LR-4 observations.

LI-RADS v2018 category and imaging features: inter-modality agreement between contrast-enhanced CT, gadoxetate disodium-enhanced MRI, and extracellular contrast-enhanced MRI.

PURPOSE: To perform an intra-individual comparison of LI-RADS category and imaging features in patients at high risk of hepatocellular carcinoma (HCC) on contrast-enhanced CT, gadoxetate disodium-enhanced MRI (EOB-MRI), and extracellular agent-enhanced MRI (ECA-MRI) and to analyze the diagnostic performance of each imaging modality. METHOD: This retrospective study included cirrhotic patients with at least one LR-3, LR-4, LR-5, LR-M or LR-TIV observation imaged with at least two imaging modalities among CT, EOB-MRI, or ECA-MRI. Two radiologists evaluated the observations using the LI-RADS v2018 diagnostic algorithm. Reference standard included pathologic confirmation and imaging criteria according to LI-RADS v2018. Imaging features were compared between different exams using the McNemar test. Inter-modality agreement was calculated by using the weighted Cohen's kappa (k) test. RESULTS: A total of 144 observations (mean size 34.0 ± 32.4 mm) in 96 patients were included. There were no significant differences in the detection of major and ancillary imaging features between the three imaging modalities. When considering all the observations, inter-modality agreement for category assignment was substantial between CT and EOB-MRI (k 0.60; 95%CI 0.44, 0.75), moderate between CT and ECA-MRI (k 0.46; 95%CI 0.22, 0.69) and substantial between EOB-MRI and ECA-MRI (k 0.72; 95%CI 0.59, 0.85). In observations smaller than 20 mm, inter-modality agreement was fair between CT and EOB-MRI (k 0.26; 95%CI 0.05, 0.47), moderate between CT and ECA-MRI (k 0.42; 95%CI -0.02, 0.88), and substantial between EOB-MRI and ECA-MRI (k 0.65; 95%CI 0.47, 0.82). ECA-MRI demonstrated the highest sensitivity (70%) and specificity (100%) when considering LR-5 as predictor of HCC. CONCLUSIONS: Inter-modality agreement between CT, ECA-MRI, and EOB-MRI decreases in observations smaller than 20 mm. ECA-MRI has the provided higher sensitivity for the diagnosis of HCC.

The American College of Radiology contrast-enhanced ultrasound Liver Imaging Reporting and Data System and its modified version in diagnosing hepatocellular carcinoma via Sonazoid: a meta-analysis.

Background: The American College of Radiology (ACR) developed the contrast-enhanced ultrasound (CEUS) Liver Imaging Reporting and Data System (LI-RADS) for pure blood contrast agents, but Sonazoid was not included. Modifications to LI-RADS have been proposed for Sonazoid. The purpose of this meta-analysis was to identify and compare the diagnostic efficacy of the two LI-RADS algorithms of Sonazoid. Methods: We searched the PubMed, MEDLINE, Web of Science, Embase, and Cochrane Library databases from databases inception to August 31, 2023, to find original studies on the ACR LI-RADS and/or modified LI-RADS algorithm with Sonazoid used as the contrast agent in patients with high-risk hepatocellular carcinoma (HCC). A bivariate random-effects model was used. Data pooling, meta-regression, and sensitivity analysis were performed for meta-analysis. The Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool was used to assess the methodological quality, and the Deeks funnel plot asymmetry test was used to evaluate the publication bias. Results: A meta-analysis of 10 studies with 1,611 observations was conducted. The pooled data for ACR LI-RADS category 5 (LR-5) and modified LR-5 were respectively as follows: pooled sensitivity, 0.70 [95% confidence interval (CI): 0.64-0.75] and 0.81 (95% CI: 0.76-0.86) (P<0.05); pooled specificity, 0.90 (95% CI: 0.82-0.94) and 0.87 (95% CI: 0.81-0.91) (P>0.05); and pooled area under the summary receiver operating characteristic curve, 0.84 and 0.91. The diagnostic performance of LI-RADS category M (LR-M) of the two algorithms was comparable. Study heterogeneity was observed. Conclusions: The results indicated that modified LR-5 algorithm demonstrated improved diagnostic sensitivity compared with the ACR LR-5 algorithm of Sonazoid, with differences observed between the different versions. Further research is needed to validate and explore the optimal diagnostic criteria for HCC using Sonazoid. Before the database search was conducted, this study was registered on PROSPERO (International Prospective Register of Systematic Reviews; CRD42023455220).

ChatGPT yields low accuracy in determining LI-RADS scores based on free-text and structured radiology reports in German language.

Background: To investigate the feasibility of the large language model (LLM) ChatGPT for classifying liver lesions according to the Liver Imaging Reporting and Data System (LI-RADS) based on MRI reports, and to compare classification performance on structured vs. unstructured reports. Methods: LI-RADS classifiable liver lesions were included from German written structured and unstructured MRI reports with report of size, location, and arterial phase contrast enhancement as minimum inclusion requirements. The findings sections of the reports were propagated to ChatGPT (GPT-3.5), which was instructed to determine LI-RADS scores for each classifiable liver lesion. Ground truth was established by two radiologists in consensus. Agreement between ground truth and ChatGPT was assessed with Cohen's kappa. Test-retest reliability was assessed by passing a subset of n = 50 lesions five times to ChatGPT, using the intraclass correlation coefficient (ICC). Results: 205 MRIs from 150 patients were included. The accuracy of ChatGPT at determining LI-RADS categories was poor (53% and 44% on unstructured and structured reports). The agreement to the ground truth was higher (k = 0.51 and k = 0.44), the mean absolute error in LI-RADS scores was lower (0.5 ± 0.5 vs. 0.6 ± 0.7, p < 0.05), and the test-retest reliability was higher (ICC = 0.81 vs. 0.50), in free-text compared to structured reports, respectively, although structured reports comprised the minimum required imaging features significantly more frequently (Chi-square test, p < 0.05). Conclusions: ChatGPT attained only low accuracy when asked to determine LI-RADS scores from liver imaging reports. The superior accuracy and consistency throughout free-text reports might relate to ChatGPT's training process. Clinical relevance statement: Our study indicates both the necessity of optimization of LLMs for structured clinical data input and the potential of LLMs for creating machine-readable labels based on large free-text radiological databases.

Radiomics-Based Machine Learning Classification Strategy for Characterization of Hepatocellular Carcinoma on Contrast-Enhanced Ultrasound in High-Risk Patients with LI-RADS Category M Nodules.

Objective Accurate differentiation within the LI-RADS category M (LR-M) between hepatocellular carcinoma (HCC) and non-HCC malignancies (mainly intrahepatic cholangiocarcinoma [CCA] and combined hepatocellular and cholangiocarcinoma [cHCC-CCA]) is an area of active investigation. We aimed to use radiomics-based machine learning classification strategy for differentiating HCC from CCA and cHCC-CCA on contrast-enhanced ultrasound (CEUS) images in high-risk patients with LR-M nodules. Methods A total of 159 high-risk patients with LR-M nodules (69 HCC and 90 CCA/cHCC-CCA) who underwent CEUS within 1 month before pathologic confirmation from January 2006 to December 2019 were retrospectively included (111 patients for training set and 48 for test set). The training set was used to build models, while the test set was used to compare models. For each observation, six CEUS images captured at predetermined time points (T1, peak enhancement after contrast injection; T2, 30 seconds; T3, 45 seconds; T4, 60 seconds; T5, 1-2 minutes; and T6, 2-3 minutes) were collected for tumor segmentation and selection of radiomics features, which included seven types of features: first-order statistics, shape (2D), gray-level co-occurrence matrix, gray-level size zone matrix, gray-level run length matrix, neighboring gray tone difference matrix, and gray-level dependence matrix. Clinical data and key radiomics features were employed to develop the clinical model, radiomics signature (RS), and combined RS-clinical (RS-C) model. The RS and RS-C model were built using the machine learning framework. The diagnostic performance of these three models was calculated and compared. Results Alpha-fetoprotein (AFP), CA19-9, enhancement pattern, and time of washout were included as independent factors for clinical model (all p < 0.05). Both the RS and RS-C model performed better than the clinical model in the test set (area under the curve [AUC] of 0.698 [0.571-0.812] for clinical model, 0.903 [0.830-0.970] for RS, and 0.912 [0.838-0.977] for the RS-C model; both p < 0.05). Conclusions Radiomics-based machine learning classifiers may be competent for differentiating HCC from CCA and cHCC-CCA in high-risk patients with LR-M nodules.

Nodule-in-nodule architecture of hepatocellular carcinomas: enhancement patterns in the hepatobiliary phase and pathological features.

PURPOSE: This study aimed to evaluate the enhancement patterns in the hepatobiliary phase (HBP) and pathological features of nodule-in-nodule-type hepatocellular carcinoma (NIN-HCC) patients. METHODS: In this single-institution retrospective study, 27 consecutive cirrhosis patients with 29 histologically confirmed NIN-HCCs who underwent preoperative examination via Gd-EOB-DTPA-enhanced MRI were enrolled from January 2016 to September 2023. Two blinded radiologists assessed the imaging features of both the inner and outer nodules in NIN-HCCs to reach a consensus on the Liver Imaging Reporting & Data System (LI-RADS) categories of the lesions. Based on the different enhancement patterns of the inner and outer nodules in the HBP, NIN-HCCs were classified into different groups and further divided into different types. Imaging features and LI-RADS categories were subsequently compared among the groups. Pathological findings for NIN-HCCs were also evaluated. RESULTS: Among 29 NIN-HCCs, all inner nodules showed hypervascularity, with a maximum diameter of 13.2 ± 5.5 mm; 51.7% (15/29) showed "wash-in with washout" enhancement; and 48.3% (14/29) showed "wash-in without washout" enhancement. All outer nodules showed hypovascularity, with a maximum diameter of 25.6 ± 7.3 mm, and 51.9% (14/29) showed a washout appearance on PVP. Among all the lesions, the maximum diameter was 27.5 ± 6.8 mm; 12 (41.4%) lesions were LR-4, and 17 (58.6%) lesions were LR-5. NIN-HCCs were classified into hypointense (62.1%, 18/29) and isointense (37.9%, 11/29) groups based on the signal intensity of the outer nodules in the HBP. In the hypointense group, 2 (6.9%) of the inner nodules were hypointense (type A), 11 (37.9%) were isointense (type B), and 5 (17.2%) were hyperintense (type C) compared to the background hypointense outer nodules. In the isointense group, 9 (31.0%) of the inner nodules were hypointense (type D), 2 (6.9%) were isointense (type E), and no (0%) was hyperintense (type F) compared to the background isointense outer nodules. There were no significant differences in the diameter, dynamic enhancement patterns of the inner or outer nodules, or LI-RADS scores of the lesions between the hypointense group and the isointense group (all P > 0.05). Histologically, the inner nodules of NIN-HCCs were mainly composed of moderately differentiated HCC (75.9% 22/29), whereas the outer nodules consisted of either well-differentiated HCC or high-grade dysplastic nodules (HGDNs). CONCLUSIONS: NIN-HCCs exhibit specific MRI findings closely associated with their pathological features. The spectrum of HBP enhancement patterns provides valuable insights into the underlying cell biological mechanisms of these lesions. NIN-HCC subtypes may be used as a morphologic marker in the early stage of multistep hepatocarcinogenesis.

Evaluation of the Contrast-Enhanced Ultrasound Nonradiation Treatment Response Assessment LI-RADS v2024 Using Data From a Multi-Center Transarterial Chemoembolization Study.

RATIONALE AND OBJECTIVE: Hepatocellular carcinoma (HCC) locoregional treatment response is commonly evaluated using the Modified Response Evaluation Criteria in Solid Tumors and the American College of Radiology (ACR) Liver Reporting and Data System (LI-RADS) Treatment Response Assessment (TRA) for MRI/CT. This study aims to evaluate the diagnostic performance of the new ACR contrast-enhanced ultrasound (CEUS) Nonradiation TRA LI-RADS v2024 in HCC treated with transarterial chemoembolization (TACE). MATERIALS AND METHODS: This retrospective observational study included 87 patients treated with TACE from a previously reported cohort. At 15- and 30-days post-treatment, 68 and 72 HCC lesions were evaluated. Three blinded radiologists with different levels of CEUS experience interpreted the images independently. According to CEUS Nonradiation TRA LI-RADSv2024, both intralesional and perilesional tumor viability were evaluated and final TRA categories were as follows: TR-Nonviable, TR-Equivocal, and TR-Viable. The reference standard used was a composite of histology and imaging. RESULTS: 140 HCC lesions were analyzed. At 15 days post-treatment, the sensitivity (SN), specificity (SP), positive predictive value (PPV), negative predictive value (NPV), and accuracy of TR-Viable classification ranged from 72.5-94.3%, 72.2-86.4%, 86.8-91.4%, 65.6-86.7%, 76.9-86.8%, respectively. At 30 days post-treatment, the SN, PPV, and NPV of TR-Viable classification decreased, ranging from 65.9-84.2%, 85.7-90.6%, and 59.5-73.9%, respectively, while the SP increased, ranging from 80.0-88.0%. Kappa values ranged from 0.557-0.730, indicating moderate to substantial agreement. CONCLUSION: CEUS Nonradiation TRA LI-RADS is a reliable tool for the detection of viable tumors in lesions treated with TACE and demonstrates reproducibility across readers.

Clinical impact of CEUS on non-characterizable observations and observations with intermediate probability of malignancy on CT/MRI in patients at risk for HCC.

BACKGROUND: Hepatocellular carcinoma (HCC) is a unique cancer allowing tumor diagnosis with identification of definitive patterns of enhancement on contrast-enhanced imaging, avoiding invasive biopsy. However, it is still unclear to what extent Contrast-Enhanced Ultrasound (CEUS) is a clinically useful additional step when Computed tomography (CT) or Magnetic resonance imaging (MRI) are inconclusive. METHODS: A prospective international multicenter validation study for CEUS Liver Imaging Reporting and Data System (LI-RADS) was conducted between January 2018 and August 2021. 646 patients at risk for HCC with focal liver lesions were enrolled. CEUS was performed using an intravenous ultrasound contrast agent within 4 weeks of CT/MRI. Liver nodules were categorized based on LI-RADS (LR) criteria. Histology or one-year follow-up CT/MRI imaging results were used as the reference standard. The diagnostic performance of CEUS was evaluated for inconclusive CT/MRI scan in two scenarios for which the AASLD recommends repeat imaging or imaging follow-up: observations deemed non-characterizable (LR-NC) or with indeterminate probability of malignancy (LR-3). RESULTS: 75 observations on CT or MRI were categorized as LR-3 (n = 54) or LR-NC (n = 21) CEUS recategorization of such observations into a different LR category (namely, into one among LR-1, LR-2, LR-5, LR-M, or LR-TIV) resulted in management recommendation changes in 33.3% (25/75) and in all but one (96.0%, 24/25) observation, the new management recommendations were correct. CONCLUSION: CEUS LI-RADS resulted in management recommendations change in substantial number of liver observations with initial indeterminate CT/MRI characterization, identifying both non-malignant lesions and HCC, potentially accelerating the diagnostic process and alleviating the need for biopsy or follow-up imaging. CLINICALTRIALS: gov number, NCT03318380.

Comparison of the Ability of Gadobenate Dimeglumine and Gadolinium Ethoxybenzyl Dimeglumine to Display the major Features for Noninvasively Diagnosing Hepatocellular Carcinoma According to the LI-RADS 2018v.

OBJECTIVE: To compare the ability of gadolinium ethoxybenzyl dimeglumine (Gd-EOB-DTPA) and gadobenate dimeglumine (Gd-BOPTA) to display the 3 major features recommended by the Liver Imaging Reporting and Data System (LI-RADS 2018v) for diagnosing hepatocellular carcinoma (HCC). MATERIALS AND METHODS: In this retrospective study, we included 98 HCC lesions that were scanned with either Gd-EOB-DTPA-MR or Gd-BOPTA-M.For each lesion, we collected multiple variables, including size and enhancement pattern in the arterial phase (AP), portal venous phase (PVP), transitional phase (TP), delayed phase (DP), and hepatobiliary phase (HBP). The lesion-to-liver contrast (LLC) was measured and calculated for each phase and then compared between the 2 contrast agents. A P value < .05 was considered statistically significant. The display efficiency of the LLC between Gd-BOPTA and Gd-EOB-DTPA for HCC features was evaluated by receiver operating characteristic (ROC) curve analysis. RESULTS: Between Gd-BOPTA and Gd-EOB-DTPA, significant differences were observed regarding the display efficiency for capsule enhancement and the LLC in the AP/PVP/DP (P < .05), but there was no significant difference regarding the LLC in the TP/HBP. Both Gd-BOPTA and Gd-EOB-DTPA had good display efficiency in each phase (AUCmin > 0.750). When conducting a total evaluation of the combined data across the 5 phases, the display efficiency was excellent (AUC > 0.950). CONCLUSION: Gd-BOPTA and Gd-EOB-DTPA are liver-specific contrast agents widely used in clinical practice. They have their own characteristics in displaying the 3 main signs of HCC. For accurate noninvasive diagnosis, the choice of agent should be made according to the specific situation.

Quantitative radiomics and qualitative LI-RADS imaging descriptors for non-invasive assessment of ß-catenin mutation status in hepatocellular carcinoma.

PURPOSE: Gain-of-function mutations in CTNNB1, gene encoding for ß-catenin, are observed in 25-30% of hepatocellular carcinomas (HCCs). Recent studies have shown ß-catenin activation to have distinct roles in HCC susceptibility to mTOR inhibitors and resistance to immunotherapy. Our goal was to develop and test a computational imaging-based model to non-invasively assess ß-catenin activation in HCC, since liver biopsies are often not done due to risk of complications. METHODS: This IRB-approved retrospective study included 134 subjects with pathologically proven HCC and available ß-catenin activation status, who also had either CT or MR imaging of the liver performed within 1 year of histological assessment. For qualitative descriptors, experienced radiologists assessed the presence of imaging features listed in LI-RADS v2018. For quantitative analysis, a single biopsy proven tumor underwent a 3D segmentation and radiomics features were extracted. We developed prediction models to assess the ß-catenin activation in HCC using both qualitative and quantitative descriptors. RESULTS: There were 41 cases (31%) with ß-catenin mutation and 93 cases (69%) without. The model's AUC was 0.70 (95% CI 0.60, 0.79) using radiomics features and 0.64 (0.52, 0.74; p = 0.468) using qualitative descriptors. However, when combined, the AUC increased to 0.88 (0.80, 0.92; p = 0.009). Among the LI-RADS descriptors, the presence of a nodule-in-nodule showed a significant association with ß-catenin mutations (p = 0.015). Additionally, 88 radiomics features exhibited a significant association (p < 0.05) with ß-catenin mutations. CONCLUSION: Combination of LI-RADS descriptors and CT/MRI-derived radiomics determine ß-catenin activation status in HCC with high confidence, making precision medicine a possibility.

Intra-individual comparison of Sonazoid contrast-enhanced ultrasound and SonoVue contrast-enhanced ultrasound in diagnosing hepatocellular carcinoma.

PURPOSE: To assess whether the diagnostic performance of Sonazoid contrast-enhanced ultrasound (SZUS) is non-inferior to that of SonoVue contrast-enhanced ultrasound (SVUS) in diagnosing hepatocellular carcinoma (HCC) in individuals with high risk. MATERIALS AND METHODS: This prospective study was conducted from October 2020 to May 2022 and included participants with a high risk of HCC who underwent SZUS and SVUS. All lesions were confirmed by clinical or pathological diagnosis. Each nodule was classified according to the Contrast-Enhanced Ultrasound Liver Imaging Reporting and Data System version 2017 (CEUS LI-RADS v2017) for SVUS and SZUS and the modified CEUS LI-RADS (using Kupffer phase defect instead of late and mild washout) for SZUS. The diagnostic performance of both two modalities for all observations was compared. Analysis of the vascular phase and Kupffer phase imaging characteristics of CEUS was performed. RESULTS: One hundred and fifteen focal liver lesions from 113 patients (94 HCCs, 12 non-HCC malignancies, and 9 benign lesions) were analysed. According to CEUS LI-RADS (v2017), SVUS and SZUS showed similar sensitivity (71.3% vs. 72.3%) and specificity (85.7% vs. 81.0%) in HCC diagnosis. However, the modified CEUS LI-RADS did not significantly improve the diagnostic efficacy of Sonazoid compared to CEUS LI-RADS v2017, having equivalent sensitivity (73.4% vs. 72.3%) and specificity (81.0% vs. 81.0%). The agreement between SVUS and SZUS for all observations was 0.610 (95% CI 0.475, 0.745), while for HCCs it was 0.452 (95% CI 0.257, 0.647). CONCLUSION: Using LI-RADS v2017, SZUS and SVUS showed non-inferior efficacy in evaluating HCC lesions. In addition, adding Kupffer phase defects to SZUS does not notably improve its diagnostic efficacy.

Using GPT-4 for LI-RADS feature extraction and categorization with multilingual free-text reports.

BACKGROUND AND AIMS: The Liver Imaging Reporting and Data System (LI-RADS) offers a standardized approach for imaging hepatocellular carcinoma. However, the diverse styles and structures of radiology reports complicate automatic data extraction. Large language models hold the potential for structured data extraction from free-text reports. Our objective was to evaluate the performance of Generative Pre-trained Transformer (GPT)-4 in extracting LI-RADS features and categories from free-text liver magnetic resonance imaging (MRI) reports. METHODS: Three radiologists generated 160 fictitious free-text liver MRI reports written in Korean and English, simulating real-world practice. Of these, 20 were used for prompt engineering, and 140 formed the internal test cohort. Seventy-two genuine reports, authored by 17 radiologists were collected and de-identified for the external test cohort. LI-RADS features were extracted using GPT-4, with a Python script calculating categories. Accuracies in each test cohort were compared. RESULTS: On the external test, the accuracy for the extraction of major LI-RADS features, which encompass size, nonrim arterial phase hyperenhancement, nonperipheral 'washout', enhancing 'capsule' and threshold growth, ranged from .92 to .99. For the rest of the LI-RADS features, the accuracy ranged from .86 to .97. For the LI-RADS category, the model showed an accuracy of .85 (95% CI: .76, .93). CONCLUSIONS: GPT-4 shows promise in extracting LI-RADS features, yet further refinement of its prompting strategy and advancements in its neural network architecture are crucial for reliable use in processing complex real-world MRI reports.

LI-RADS version 2018 treatment response algorithm on extracellular contrast-enhanced MRI in patients treated with transarterial chemoembolization for hepatocellular carcinoma: diagnostic performance and the added value of ancillary features.

BACKGROUND: The Liver Imaging Reporting and Data System (LI-RADS) Treatment Response Algorithm (TRA) (LI-RADS TRA) is used for assessing response of HCC to locoregional therapy (LRT), however, the value of ancillary features (AFs) for TACE-treated HCCs has not been extensively investigated on extracellular agent MRI (ECA-MRI). PURPOSE: To evaluate the diagnostic performance of LI-RADS v2018 TRA on ECA-MRI for HCC treated with transarterial chemoembolization (TACE) and the value of ancillary features. METHODS: This retrospective study included patients who underwent TACE for HCC and then followed by hepatic surgery between January 2019 and June 2023 with both pre- and post-TACE contrast-enhanced MRI available. Two radiologists independently evaluated the post-treated lesions on MRI using LI-RADS treatment response (TR) (LR-TR) algorithm and modified LR-TR (mLR-TR) algorithm in which ancillary features (restricted diffusion and intermediate T2-weighted hyperintensity) were added, respectively. Lesions were categorized as complete pathologic necrosis (100%, CPN) and non-complete pathologic necrosis (< 100%, non-CPN) on the basis of surgical pathology. The diagnostic performance in predicting viable and non-viable tumors based on LR-TR and mLR-TR algorithms was compared using the McNemar test. Interreader agreement was calculated by using Cohen's weighted and unweighted κ. RESULTS: A total of 61 patients [mean age 59 years ± 10 (standard deviation); 47 men] with 79 lesions (57 pathologically viable) were included. For non-CPN prediction, the sensitivity, specificity of LR-TR viable and mLR-TR viable category were 75% (43 of 57), 82% (18 of 22) and 88% (50 of 57), 77% (17 of 22), respectively, the sensitivity of mLR-TR was significantly higher than that of LR-TR (P = 0.016) without difference in specificity (P = 1.000). Interreader agreement for LR-TR and mLR-TR category was moderate (k = 0.50, 95% confidence interval 0.33, 0.67, k = 0.42, 95% confidence interval 0.20, 0.63). The sensitivity of both LR-TR and mLR-TR algorithms in predicting viable tumors between conventional TACE (cTACE) and drug-eluting beads TACE (DEB-TACE) did not have significant difference (cTACE: 76%, 89% vs. DEB-TACE: 73%, 82%). CONCLUSIONS: On ECA-MRI, applying ancillary features to LI-RADS v2018 TRA can improve the sensitivity in predicting pathologic tumor viability in patients treated with TACE for hepatocellular carcinoma with no significant difference in specificity.

Deep learning-based arterial subtraction images improve the detection of LR-TR algorithm for viable HCC on extracellular agents-enhanced MRI.

PURPOSE: To determine the role of deep learning-based arterial subtraction images in viability assessment on extracellular agents-enhanced MRI using LR-TR algorithm. METHODS: Patients diagnosed with HCC who underwent locoregional therapy were retrospectively collected. We constructed a deep learning-based subtraction model and automatically generated arterial subtraction images. Two radiologists evaluated LR-TR category on ordinary images and then evaluated again on ordinary images plus arterial subtraction images after a 2-month washout period. The reference standard for viability was tumor stain on the digital subtraction hepatic angiography within 1 month after MRI. RESULTS: 286 observations of 105 patients were ultimately enrolled. 157 observations were viable and 129 observations were nonviable according to the reference standard. The sensitivity and accuracy of LR-TR algorithm for detecting viable HCC significantly increased with the application of arterial subtraction images (87.9% vs. 67.5%, p < 0.001; 86.4% vs. 75.9%, p < 0.001). And the specificity slightly decreased without significant difference when the arterial subtraction images were added (84.5% vs. 86.0%, p = 0.687). The AUC of LR-TR algorithm significantly increased with the addition of arterial subtraction images (0.862 vs. 0.768, p < 0.001). The arterial subtraction images also improved inter-reader agreement (0.857 vs. 0.727). CONCLUSION: Extended application of deep learning-based arterial subtraction images on extracellular agents-enhanced MRI can increase the sensitivity of LR-TR algorithm for detecting viable HCC without significant change in specificity.

Compared with SonoVue® LR-5, Sonazoid® modified LR-5 has better diagnostic sensitivity for hepatocellular carcinoma: a systematic review and meta-analysis.

Background: The contrast-enhanced ultrasound (CEUS) liver imaging reporting and data system (LI-RADS) is a standardized system for reporting liver nodules in patients at risk of developing hepatocellular carcinoma (HCC) and is only recommended for pure blood pool agents such as SonoVue®. A modified LI-RADS was proposed for Sonazoid®, a Kupffer cell-specific contrast agent. This meta-analysis was conducted to compare the diagnostic efficiency of the CEUS LI-RADS for SonoVue® and the modified LI-RADS for Sonazoid®. Methods: The PubMed, Medline, Web of Science, Embase, and Cochrane Library databases were systematically searched to retrieve studies on the diagnostic efficiency of the CEUS LI-RADS algorithms in diagnosing HCC using SonoVue® and/or Sonazoid® from January 2016 to June 2023. Histopathology or imaging follow-up served as the reference standards. Only articles published in English on retrospective or prospective studies with full reports were included in the meta-analysis. A bivariate random-effects model was used. Data pooling, meta-regression, and sensitivity analysis were performed for the meta-analysis. Deeks' funnel plot asymmetry test was used to evaluate publication bias, and the QUADAS-2 tool was used to assess the methodological quality of eligible studies. Results: In total, 26 studies comprising 8,495 patients with 9,244 lesions were included in the meta-analysis. The pooled data results for SonoVue® LI-RADS category 5 (LR-5) and Sonazoid® modified LR-5 were as follows: pooled sensitivity: 0.68 [95% confidence interval (CI): 0.64-0.73, I2=89.20%; P<0.01] and 0.82 (95% CI: 0.74-0.87, I2=85.39%; P<0.01) (P<0.05); pooled specificity: 0.93 (95% CI: 0.90-0.96, I2=86.52%; P<0.01) and 0.86 (95% CI: 0.79-0.91, I2=59.91%; P=0.01) (P<0.05); pooled area under the curve (AUC): 0.86 (95% CI: 0.82-0.89) and 0.91 (95% CI: 0.88-0.93) (P<0.05), respectively. The meta-regression analysis revealed that the study design, subject enrollment method, and reference standard contributed to the heterogeneity of SonoVue® LR-5, and the number of lesions was a source of heterogeneity for Sonazoid® modified LR-5. The diagnostic performance of the LI-RADS category M (LR-M) algorithms of SonoVue® and Sonazoid® was comparable. Conclusions: The Sonazoid® modified LR-5 algorithm had a higher diagnostic sensitivity, lower specificity, and higher AUC than SonoVue® LR-5.

The efficacy of modified contrast-enhanced ultrasound Liver Imaging Reporting and Data System (CEUS LI-RADS) using Sonazoid in diagnosis of hepatocellular carcinoma: a systematic review and meta-analysis.

Background: The contrast-enhanced ultrasound Liver Imaging Reporting and Data System (CEUS LI-RADS) is an algorithm for the diagnosis of hepatocellular carcinoma (HCC) in high-risk populations. Previous studies have shown the algorithm to have high specificity and moderate sensitivity. Nevertheless, it is designated for utilization solely with blood pool contrast agents. Sonazoid, a contrast agent that combines blood pools and Kupffer cells properties, has recently gained approval for marketing in an increased number of countries. Enhanced sensitivity in diagnosing HCC may be achieved through the distinctive Kupffer phase (KP) exhibited by Sonazoid. Certain academics have suggested the modified CEUS LI-RADS using Sonazoid. The main criteria of mild and late (≥60 seconds) washout in CEUS LI-RADS LR-5 were replaced by KP (>10 minutes) defects as the primary criteria. The purpose of this research was to evaluate the effectiveness of the modified CEUS LI-RADS using Sonazoid in diagnosing HCC. Methods: Original studies on Sonazoid and CEUS LI-RADS were searched in the PubMed, Embase, Cochrane Library, and Web of Science databases until 13 July 2023, with no restrictions on language. We enrolled studies that applied Sonazoid for CEUS in patients at high risk of HCC and modified CEUS LI-RADS for the diagnosis of intrahepatic nodules. Meta-analyses, evaluations, case studies, correspondences, remarks, and summaries of conferences were excluded. Additionally, studies that fell outside the scope of this study and contained data on the same patients were also excluded. We evaluated the quality of research by employing the Quality Assessment of Diagnostic Accuracy Studies 2 (QUADAS-2) tool. A bivariate mixed effects model was utilized to conduct a meta-analysis, summarizing the sensitivity and specificity in the diagnosis of HCC. The investigation of potential factors contributing to study heterogeneity was conducted using meta-regression analysis. Results: Out of the 103 studies screened, 6 studies (835 lesions) were included in the final results. Modified CEUS LR-5 exhibited a sensitivity of 0.77 [95% confidence interval (CI): 0.70-0.82; I2=71.98%; P=0.00] and a specificity of 0.88 (95% CI: 0.83-0.92; I2=0.00; P=0.47) for HCC diagnosis, with heterogeneity in sensitivity. The presence of heterogeneity in the study was found to have a significant association with factors such as the study design, the number of image reviewers, the proportion of cirrhosis, the proportion of other non-HCC malignancies (OM) cases, and the type of reference standard (P≤0.05). Conclusions: The modified CEUS LI-RADS LR-5 categorization demonstrates a reasonable level of sensitivity 0.77, but an insufficient level of specificity 0.88 when diagnosing HCC. KP defects cannot be used as a primary feature in the diagnosis of HCC by CEUS LI-RADS, perhaps as an ancillary feature.

Enhancement Pattern Mapping for Early Detection of Hepatocellular Carcinoma in Patients with Cirrhosis.

Background and Aims: Limited methods exist to accurately characterize the risk of malignant progression of liver lesions. Enhancement pattern mapping (EPM) measures voxel-based root mean square deviation (RMSD) of parenchyma and the contrast-to-noise (CNR) ratio enhances in malignant lesions. This study investigates the utilization of EPM to differentiate between HCC versus cirrhotic parenchyma with and without benign lesions. Methods: Patients with cirrhosis undergoing MRI surveillance were studied prospectively. Cases (n=48) were defined as patients with LI-RADS 3 and 4 lesions who developed HCC during surveillance. Controls (n=99) were patients with and without LI-RADS 3 and 4 lesions who did not develop HCC. Manual and automated EPM signals of liver parenchyma between cases and controls were quantitatively validated on an independent patient set using cross validation with manual methods avoiding parenchyma with artifacts or blood vessels. Results: With manual EPM, RMSD of 0.37 was identified as a cutoff for distinguishing lesions that progress to HCC from background parenchyma with and without lesions on pre-diagnostic scans (median time interval 6.8 months) with an area under the curve (AUC) of 0.83 (CI: 0.73-0.94) and a sensitivity, specificity, and accuracy of 0.65, 0.97, and 0.89, respectively. At the time of diagnostic scans, a sensitivity, specificity, and accuracy of 0.79, 0.93, and 0.88 were achieved with manual EPM with an AUC of 0.89 (CI: 0.82-0.96). EPM RMSD signals of background parenchyma that did not progress to HCC in cases and controls were similar (case EPM: 0.22 ± 0.08, control EPM: 0.22 ± 0.09, p=0.8). Automated EPM produced similar quantitative results and performance. Conclusion: With manual EPM, a cutoff of 0.37 identifies quantifiable differences between HCC cases and controls approximately six months prior to diagnosis of HCC with an accuracy of 89%.

Current surveillance and diagnostic methods in hepatocellular carcinoma are suboptimal. Enhancement pattern mapping is an imaging technique that quantifies lesion signals and may be useful in diagnostic and surveillance methods. Enhancement pattern mapping describes quantifiable differences between malignant and benign liver tissue on contrast-enhanced MRI. It amplifies lesion signal and distinguishes malignancy in a surveillance population. The novel imaging technique was investigated at single institution and analyzed lesions compared to cirrhotic parenchyma. Future efforts will include further risk stratification across LI-RADS group categories. The results provide evidence that enhancement pattern mapping uses available imaging data to distinguish hepatocellular carcinoma from non-cancerous parenchyma with and without benign lesions on scans six months prior to diagnosis with standard MRI. The technique introduces a prospective modality to improve diagnostic accuracy and early detection with the goal of improving clinical outcomes.

Optimizing LI-RADS: ancillary features screened from LR-3/4 categories can improve the diagnosis of HCC on MRI.

OBJECTIVE: To determine the high-efficiency ancillary features (AFs) screened from LR-3/4 lesions and the HCC/non-HCC group and the diagnostic performance of LR3/4 observations. MATERIALS AND METHODS: We retrospectively analyzed a total of 460 patients (with 473 nodules) classified into LR-3-LR-5 categories, including 311 cases of hepatocellular carcinoma (HCC), 6 cases of non-HCC malignant tumors, and 156 cases of benign lesions. Two faculty abdominal radiologists with experience in hepatic imaging reviewed and recorded the major features (MFs) and AFs of the Liver Imaging Reporting and Data System (LI-RADS). The frequency of the features and diagnostic performance were calculated with a logistic regression model. After applying the above AFs to LR-3/LR-4 observations, the sensitivity and specificity for HCC were compared. RESULTS: The average age of all patients was 54.24 ± 11.32 years, and the biochemical indicators ALT (P = 0.044), TBIL (P = 0.000), PLT (P = 0.004), AFP (P = 0.000) and Child‒Pugh class were significantly higher in the HCC group. MFs, mild-moderate T2 hyperintensity, restricted diffusion and AFs favoring HCC in addition to nodule-in-nodule appearance were common in the HCC group and LR-5 category. AFs screened from the HCC/non-HCC group (AF-HCC) were mild-moderate T2 hyperintensity, restricted diffusion, TP hypointensity, marked T2 hyperintensity and HBP isointensity (P = 0.005, < 0.001, = 0. 032, p < 0.001, = 0.013), and the AFs screened from LR-3/4 lesions (AF-LR) were restricted diffusion, mosaic architecture, fat in mass, marked T2 hyperintensity and HBP isointensity (P < 0.001, = 0.020, = 0.036, < 0.001, = 0.016), which were not exactly the same. After applying AF-HCC and AF-LR to LR-3 and LR-4 observations in HCC group and Non-HCC group, After the above grades changed, the diagnostic sensitivity for HCC were 84.96% using AF-HCC and 85.71% using AF-LR, the specificity were 89.26% using AF-HCC and 90.60% using AF-LR, which made a significant difference (P = 0.000). And the kappa value for the two methods of AF-HCC and AF-LR were 0.695, reaching a substantial agreement. CONCLUSION: When adjusting for LR-3/LR-4 lesions, the screened AFs with high diagnostic ability can be used to optimize LI-RADS v2018; among them, AF-LR is recommended for better diagnostic capabilities.