Increased interreader agreement in diagnosis of hepatocellular carcinoma using an adapted LI-RADS algorithm

Evidence Supporting Diagnostic Value of Liver Imaging Reporting and Data System for CT- and MR Imaging-based Diagnosis of Hepatocellular Carcinoma: A Systematic Review and Meta-analysis

Background

Based on the Liver Imaging Data and Reporting System (LI-RADS) guidelines, Hepatocellular Carcinoma (HCC) can be diagnosed using imaging criteria in patients at risk of HCC.

Objective

This study aimed to assess the diagnostic value of LI-RADS in high-risk patients with HCC.

Material and Methods

This systematic review is conducted on international databases, including Google Scholar, Web of Science, PubMed, Embase, PROQUEST, and Cochrane Library, with appropriate keywords. Using the binomial distribution formula, the variance of each study was calculated, and all the data were analyzed using STATA version 16. The pooled sensitivity and specificity were determined using a random-effects meta-analysis approach. Also, we used the chi-squared test and I2 index to calculate heterogeneity among studies, and Funnel plots and Egger tests were used for evaluating publication bias.

Results

The pooled sensitivity was estimated at 0.80 (95% CI 0.76-0.84). According to different types of Liver Imaging Reporting and Data Systems (LI-RADS), the highest pooled sensitivity was in version 2018 (0.83 (95% CI 0.79-0.87) (I2: 80.6%, P of chi 2 test for heterogeneity <0.001 and T2: 0.001). The pooled specificity was estimated as 0.89 (95% CI 0.87-0.92). According to different types of LI-RADS, the highest pooled specificity was in version 2014 (93.0 (95% CI 89.0-96.0) (I2: 81.7%, P of chi 2 test for heterogeneity <0.001 and T2: 0.001).

Conclusion

LI-RADS can assist radiologists in achieving the required sensitivity and specificity in high-risk patients suspected to have HCC. Therefore, this strategy can serve as an appropriate tool for identifying HCC.

Individual Participant Data Meta-Analysis of LR-5 in LI-RADS Version 2018 versus Revised LI-RADS for Hepatocellular Carcinoma Diagnosis

Background A simplification of the Liver Imaging Reporting and Data System (LI-RADS) version 2018 (v2018), revised LI-RADS (rLI-RADS), has been proposed for imaging-based diagnosis of hepatocellular carcinoma (HCC). Single-site data suggest that rLI-RADS category 5 (rLR-5) improves sensitivity while maintaining positive predictive value (PPV) of the LI-RADS v2018 category 5 (LR-5), which indicates definite HCC. Purpose To compare the diagnostic performance of LI-RADS v2018 and rLI-RADS in a multicenter data set of patients at risk for HCC by performing an individual patient data meta-analysis. Materials and Methods Multiple databases were searched for studies published from January 2014 to January 2022 that evaluated the diagnostic performance of any version of LI-RADS at CT or MRI for diagnosing HCC. An individual patient data meta-analysis method was applied to observations from the identified studies. Quality Assessment of Diagnostic Accuracy Studies version 2 was applied to determine study risk of bias. Observations were categorized according to major features and either LI-RADS v2018 or rLI-RADS assignments. Diagnostic accuracies of category 5 for each system were calculated using generalized linear mixed models and compared using the likelihood ratio test for sensitivity and the Wald test for PPV. Results Twenty-four studies, including 3840 patients and 4727 observations, were analyzed. The median observation size was 19 mm (IQR, 11-30 mm). rLR-5 showed higher sensitivity compared with LR-5 (70.6% [95% CI: 60.7, 78.9] vs 61.3% [95% CI: 45.9, 74.7]; P < .001), with similar PPV (90.7% vs 92.3%; P = .55). In studies with low risk of bias (n = 4; 1031 observations), rLR-5 also achieved a higher sensitivity than LR-5 (72.3% [95% CI: 63.9, 80.1] vs 66.9% [95% CI: 58.2, 74.5]; P = .02), with similar PPV (83.1% vs 88.7%; P = .47). Conclusion rLR-5 achieved a higher sensitivity for identifying HCC than LR-5 while maintaining a comparable PPV at 90% or more, matching the results presented in the original rLI-RADS study. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Sirlin and Chernyak in this issue.

A comparative study between deep learning and radiomics models in grading liver tumors using hepatobiliary phase contrast-enhanced MR images

Purpose

To compare a deep learning model with a radiomics model in differentiating high-grade (LR-3, LR-4, LR-5) liver imaging reporting and data system (LI-RADS) liver tumors from low-grade (LR-1, LR-2) LI-RADS tumors based on the contrast-enhanced magnetic resonance images.

Methods

Magnetic resonance imaging scans of 361 suspected hepatocellular carcinoma patients were retrospectively reviewed. Lesion volume segmentation was manually performed by two radiologists, resulting in 426 lesions from the training set and 83 lesions from the test set. The radiomics model was constructed using a support vector machine (SVM) with pre-defined features, which was first selected using Chi-square test, followed by refining using binary least absolute shrinkage and selection operator (LASSO) regression. The deep learning model was established based on the DenseNet. Performance of the models was quantified by area under the receiver-operating characteristic curve (AUC), accuracy, sensitivity, specificity and F1-score.

Results

A set of 8 most informative features was selected from 1049 features to train the SVM classifier. The AUCs of the radiomics model were 0.857 (95% confidence interval [CI] 0.816–0.888) for the training set and 0.879 (95% CI 0.779–0.935) for the test set. The deep learning method achieved AUCs of 0.838 (95% CI 0.799–0.871) for the training set and 0.717 (95% CI 0.601–0.814) for the test set. The performance difference between these two models was assessed by t-test, which showed the results in both training and test sets were statistically significant.

Conclusion

The deep learning based model can be trained end-to-end with little extra domain knowledge, while the radiomics model requires complex feature selection. However, this process makes the radiomics model achieve better performance in this study with smaller computational cost and more potential on model interpretability.

Interrater reliability and agreement of the liver imaging reporting and data system (LI-RADS) v2018 for the evaluation of hepatic lesions

Purpose

The liver imaging reporting and data system (LI-RADS) is a structured reporting system that categorizes hepatic observations according to major imaging features and lesion size, with an optional ancillary features contribution. This study aimed to evaluate inter-reader agreement of dynamic magnetic resonance imaging (MRI) using LI-RADS v2018 lexicon.

Material and methods

Forty-nine patients with 69 hepatic observations were included in our study. The major and ancillary features of each hepatic observation were evaluated by 2 radiologists using LI-RADS v2018, and the interreader agreement was allocated.

Results

The inter-reader agreement of major LI-RADS features was substantial; κ of non-rim arterial hyperenhancement, non-peripheral washout appearance, and enhancing capsule was 0.796, 0.799, and 0.772 (p < 0.001), respectively. The agreement of the final LI-RADS category was substantial with κ = 0.651 (p < 0.001), and weighted κ = 0.786 (p < 0.001). The inter-reader agreement of the ancillary features was substantial to almost perfect (k range from 0.718 to 1; p < 0.001). An almost perfect correlation was noted for the hepatic lesion size measurement with ICC = 0.977 (p < 0.001).

Conclusions

The major and ancillary features of the LI-RADS v2018, as well as the final category and lesions size, have substantial to almost perfect inter-reader agreement.

Saliency-based 3D convolutional neural network for categorising common focal liver lesions on multisequence MRI

Background

The imaging features of focal liver lesions (FLLs) are diverse and complex. Diagnosing FLLs with imaging alone remains challenging. We developed and validated an interpretable deep learning model for the classification of seven categories of FLLs on multisequence MRI and compared the differential diagnosis between the proposed model and radiologists.

Methods

In all, 557 lesions examined by multisequence MRI were utilised in this retrospective study and divided into training–validation (n = 444) and test (n = 113) datasets. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate the performance of the model. The accuracy and confusion matrix of the model and individual radiologists were compared. Saliency maps were generated to highlight the activation region based on the model perspective.

Results

The AUC of the two- and seven-way classifications of the model were 0.969 (95% CI 0.944–0.994) and from 0.919 (95% CI 0.857–0.980) to 0.999 (95% CI 0.996–1.000), respectively. The model accuracy (79.6%) of the seven-way classification was higher than that of the radiology residents (66.4%, p = 0.035) and general radiologists (73.5%, p = 0.346) but lower than that of the academic radiologists (85.4%, p = 0.291). Confusion matrices showed the sources of diagnostic errors for the model and individual radiologists for each disease. Saliency maps detected the activation regions associated with each predicted class.

Conclusion

This interpretable deep learning model showed high diagnostic performance in the differentiation of FLLs on multisequence MRI. The analysis principle contributing to the predictions can be explained via saliency maps.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13244-021-01117-z.

Progression Rates of LR-2 and LR-3 Observations on MRI to Higher LI-RADS Categories in Patients at High Risk for Hepatocellular Carcinoma: A Retrospective Study

Background: Reported rates of hepatocellular carcinoma (HCC) for LR-2 and LR-3 observations are generally greater than what may be expected based on clinical experience, possibly reflecting some studies' requirement for pathologic reference. Objective: To determine progression rates to higher LI-RADS categories of LR-2 and LR-3 observations in patients at high risk for HCC. Methods: This retrospective study included 91 patients (mean age 62 years; 64 men, 27 women) at high risk for HCC with clinically reported LR-2 (n=55) or LR-3 (n=36) observations on MRI and who also underwent follow-up CT or MRI after at least 12 months. A study coordinator annotated the location of a single LR-2 or LR-3 observation per patient, based on the clinical reports. Using LI-RADS v2018 criteria, two radiologists independently assigned LI-RADS categories on the follow-up examinations. Progression rates from LR-2 or LR-3 to higher categories were determined. A post hoc consensus review was performed of observations that progressed to LR-4 or LR-5. Subgroup analyses were performed with respect to presence or prior HCC (n=34) or a separate baseline LR-5 observation (n=12). Results: For LR-2 observations, progression to LR-4 was 0.0% (95% CI 0.0%-6.7%) and LR-5 was 3.6% (95% CI 0.4%-13.1%) for both readers. For LR-3 observations, progression to LR-4 was 22.2% (95% CI 9.6%-43.8%) and LR-5 was 11.1% (95% CI 3.0%-28.4%) for both readers. Fourteen observations progressed to LR-4 or LR-5 for both readers (post hoc analysis: no LR-2 to LR-4; two LR-2 to LR-5; eight LR-3 to LR-4; four LR-3 to LR-5). Progression rate from LR-3 to LR-5 was higher (p<.001) among patients with versus without a separate baseline LR-5 observation (100.0% vs 3.0%, respectively) for both readers; progression rate from LR-2 to LR-5 was not associated with a separate baseline LR-5 observation for either reader (p=.30). Progression rates were not different (p>.99) between patients with versus without prior HCC. Conclusion: Based on progression to LR-4 or LR-5, LR-2 and LR-3 observations showed lower progression rates than reported in studies incorporating pathology in determining progression. Clinical Impact: The findings refine understanding of the clinical significance of LR-2 and LR-3 observations.

Does Training in LI-RADS Version 2018 Improve Readers' Agreement with the Expert Consensus and Inter-reader Agreement in MRI Interpretation?

BACKGROUND

The Liver Imaging Reporting and Data System (LI-RADS) was established for noninvasive diagnosis for hepatocellular carcinoma (HCC). However, whether training can improve readers' agreement with the expert consensus and inter-reader agreement for final categories is still unclear.

PURPOSE

To explore training effectiveness on readers' agreement with the expert consensus and inter-reader agreement.

STUDY TYPE

Prospective.

SUBJECTS

Seventy lesions in 61 patients at risk of HCC undergoing liver MRI; 20 visiting scholars.

FIELD STRENGTH/SEQUENCE

1.5 T or 3 T, Dual-echo T₁ WI, Fast spin-echo T₂ WI, SE-EPI DWI, and Dynamic multiphase fast gradient-echo T₁ WI.

ASSESSMENT

Seventy lesions assigned LI-RADS categories of LR1-LR5, LR-M, and LR-TIV by three radiologists in consensus were randomly selected, with 10 cases for each category. The consensus opinion was the standard reference. The third radiologist delivered the training. Twenty readers reviewed images independently and assigned each an LI-RADS category both before and after the training.

STATISTICAL TESTS

Accuracy, sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio, receiver operating characteristic (ROC) analysis, simple and weighted kappa statistics, and Fleiss kappa statistics.

RESULTS

Before and after training: readers' AUC (areas under ROC) for LR-1-LR-5, LR-M, and LR-TIV were 0.898 vs. 0.913, 0.711 vs. 0.876, 0.747 vs. 0.860, 0.724 vs. 0.815, 0.844 vs. 0.895, 0.688 vs. 0.873, and 0.720 vs. 0.948, respectively, and all improved significantly (P < 0.05), except LR-1(P = 0.25). Inter-reader agreement between readers for LR-1-LR-5, LR-M, LR-TIV were 0.725 vs. 0.751, 0.325 vs. 0.607, 0.330 vs. 0.559, 0.284 vs. 0.488, 0.447 vs. 0.648, 0.229 vs. 0.589, and 0.362 vs. 0.852, respectively, and all increased significantly (P < 0.05). For training effectiveness on both AUC and inter-reader agreement, LR-TIV, LR-M, and LR-2 improved most, and LR-1 made the least.

DATA CONCLUSION

This study shows LI-RADS training could improve reader agreement with the expert consensus and inter-reader agreement for final categories.

LEVEL OF EVIDENCE

2 TECHNICAL EFFICACY STAGE: 2.

Diagnostic performance of LI-RADS for MRI and CT detection of HCC: A systematic review and diagnostic meta-analysis

PURPOSE

To perform a meta-analysis evaluating the diagnostic accuracy of the Liver Imaging Reporting and Data System (LI-RADS) category ≥ 3 (LI-RADS 3-5v) for detecting hepatocellular carcinoma (HCC).

METHOD

A systematic PubMed, Embase, and Web of Science electronic database search was performed for original diagnostic studies published through July 31, 2018. Statistical analysis included data pooling, forest plot construction, heterogeneity testing, meta-regression, and subgroup analyses.

RESULTS

Eighteen studies (v2011, v2014 and v2017) involving 3386 patients were included in the meta-analysis. The pooled sensitivity and specificity of LI-RADS ≥ 3 for diagnosing HCC were 0.86 (95 % confidence interval (CI): 0.78-0.91) and 0.85 (95 % CI: 0.78-0.90), respectively. The area under the curve (AUC) was 0.92 (95 % CI: 0.89-0.94). Meta-regression analysis showed that the publication year, blinding to the reference standard and the number of readers were significant factors affecting heterogeneity. In subgroup analyses, magnetic resonance imaging (MRI) demonstrated higher sensitivity (0.82 vs. 0.73) and comparable specificity (0.79 vs. 0.78) than computed tomography (CT). For HCCs ≤30 mm, LI-RADS showed lower sensitivity of 0.72 and specificity of 0.80 compared with HCC of all sizes. LR-5 showed higher sensitivity and specificity than LR-3 (sensitivity: 0.67 vs. 0.07, P = 0.02; specificity: 0.93 vs. 0.75, p < 0.001) and higher sensitivity than LR-4 (sensitivity: 0.67 vs. 0.29, P = 0.02; specificity: 0.93 vs. 0.80, p = 0.75). LR ≥ 5 had higher specificity at the cost of decreased sensitivity than LR ≥ 3 (specificity: 0.94 vs. 0.68, p < 0.001; sensitivity: 0.66 vs. 0.74, P = 0.70) and LR ≥ 4 (specificity: 0.94 vs. 0.84, p < 0.001; sensitivity: 0.66 vs. 0.74, P = 0.77).

CONCLUSIONS

LI-RADS ≥ 3 shows high diagnostic accuracy for HCCs, with a pooled sensitivity of 0.86 and specificity of 0.85. The specificity is higher for LR-5 and LR ≥ 5. However, further prospective studies on LI-RADS ≥ 3 are needed to elucidate its value for diagnosing small HCCs (≤20 mm).

Interreader Agreement of Liver Imaging Reporting and Data System on MRI: A Systematic Review and Meta-Analysis

BACKGROUND

Use of the Liver Imaging Reporting and Data System (LI-RADS) is increasing, but the reported results for interreader agreement seem quite variable.

PURPOSE

To systematically determine the interreader agreement of LI-RADS on magnetic resonance imaging (MRI) and to determine the sources of heterogeneity between the reported results.

STUDY TYPE

Systematic review and meta-analysis.

SUBJECTS

Fifteen original articles with 2968 lesions.

FIELD STRENGTH

1.5T and 3.0T.

ASSESSMENT

Two reviewers independently performed the data extraction. The reviewers identified and reviewed the original articles reporting the interreader agreement of LI-RADS using MRI.

STATISTICAL TESTS

The meta-analytic pooled intraclass correlation coefficient (ICC) for lesion size and kappa value (κ) for major features (arterial-phase hyperenhancement [APHE], nonperipheral washout [WO], enhancing capsule [EC]) and LI-RADS categorization (LR) were calculated using the random-effects model. Sensitivity analysis and meta-regression analysis were performed to explore the cause of study heterogeneity.

RESULTS

The meta-analytic pooled ICC of lesion size was 0.97 (95% confidence interval [CI], 0.94-1.00). Meta-analytic pooled κ of APHE, WO, EC, and LR were 0.72 (95% CI, 0.62-0.82), 0.69 (95% CI, 0.60-0.78), 0.66 (95% CI, 0.58-0.74), and 0.70 (95% CI, 0.56-0.85), respectively. Substantial study heterogeneity was noted in all five variables (I2  ≥ 89.1%, P < 0.001). Study design, type, and clarity of blinding review were factors that significantly influenced study heterogeneity (P ≤ 0.05).

DATA CONCLUSION

LI-RADS demonstrated overall substantial interreader agreement for major features and the category on MRI, but showed heterogeneous results between studies.

LEVEL OF EVIDENCE

3 TECHNICAL EFFICACY STAGE: 2 J. Magn. Reson. Imaging 2020;52:795-804.

Intraindividual Comparison between Gadoxetate-Enhanced Magnetic Resonance Imaging and Dynamic Computed Tomography for Characterizing Focal Hepatic Lesions: A Multicenter, Multireader Study

OBJECTIVE

To compare the diagnostic accuracy of dynamic computed tomography (CT) and gadoxetate-enhanced magnetic resonance imaging (MRI) for characterization of hepatic lesions by using the Liver Imaging Reporting and Data System (LI-RADS) in a multicenter, off-site evaluation.

MATERIALS AND METHODS

In this retrospective multicenter study, we evaluated 231 hepatic lesions (114 hepatocellular carcinomas [HCCs], 58 non-HCC malignancies, and 59 benign lesions) confirmed histologically in 217 patients with chronic liver disease who underwent both gadoxetate-enhanced MRI and dynamic CT at one of five tertiary hospitals. Four radiologists at different institutes independently reviewed all MR images first and the CT images 4 weeks later. They evaluated the major and ancillary imaging features and categorized each hepatic lesion according to the LI-RADS v2014. Diagnostic performance was calculated and compared using generalized estimating equations.

RESULTS

MRI showed higher sensitivity and accuracy than CT for diagnosing hepatic malignancies; the pooled sensitivities, specificities, and accuracies for categorizing LR-5/5V/M were 59.0% vs. 72.4% (CT vs. MRI; p < 0.001), 83.5% vs. 83.9% (p = 0.906), and 65.3% vs. 75.3% (p < 0.001), respectively. CT and MRI showed comparable capabilities for differentiating between HCC and other malignancies, with pooled accuracies of 79.9% and 82.4% for categorizing LR-M, respectively (p = 0.139).

CONCLUSION

Gadoxetate-enhanced MRI showed superior accuracy for categorizing LR-5/5V/M in hepatic malignancies in comparison with dynamic CT. Both modalities had comparable accuracies for distinguishing other malignancies from HCC.

Validation of the Liver Imaging Reporting and Data System Treatment Response Criteria After Thermal Ablation for Hepatocellular Carcinoma

Single hepatocellular carcinoma (HCC) tumors can be successfully eradicated with thermal ablation (TA). We assessed the validity of the Liver Imaging Reporting and Data System Treatment Response (LR-TR) criteria with a retrospective analysis of a single-center database of patients with small HCC tumors (<3 cm in diameter) who underwent both laparoscopic TA and liver transplantation (LT) from 2004 to 2018. Postablation MRIs were assigned LR-TR categories (nonviable, equivocal, and viable) for ablated lesions and Liver Imaging Reporting and Data System (LI-RADS) categories (probable or definite HCC) for untreated lesions. Interpretations were compared with the histopathology of the post-LT explanted liver. There were 45 patients with 81 tumors (59 ablated and 22 untreated; mean size, 2.2 cm), and 23 (39%) of the ablated tumors had viable HCC on histopathology. The sensitivity/specificity of LR-TR categories (nonviable/equivocal versus viable) of ablated tumors was 30%/99%, with a positive predictive value (PPV)/negative predictive value (NPV) of 93%/69%. The sensitivity varied with residual tumor size. The sensitivity/specificity of LI-RADS 4 and 5 diagnostic criteria at detecting new HCC was 65%/94%, respectively, with a PPV/NPV of 85%/84%. The interrater reliability (IRR) was high for LR-TR categories (90% agreement, Cohen's ĸ = 0.75) and for LI-RADS LR-4 and LR-5 diagnostic categories (91% agreement, Cohen's ĸ = 0.80). In patients with HCC <3 cm in diameter, LR-TR criteria after TA had high IRR but low sensitivity, suggesting that the LR-TR categories are precise but inaccurate. The low sensitivity may be secondary to TA's disruption in the local blood flow of the tissue, which could affect the arterial enhancement phase on MRI. Additional investigation and new technologies may be necessary to improve imaging after ablation.

Interobserver Agreement of Magnetic Resonance Imaging of Liver Imaging Reporting and Data System Version 2018

AIM

This study aimed to assess the interobserver agreement of magnetic resonance (MR) imaging of Liver Imaging Reporting and Data System version 2018 (LI-RADS v2018).

SUBJECTS AND METHODS

Retrospective analysis was done for 119 consecutive patients (77 male and 42 female) at risk of hepatocellular carcinoma who underwent dynamic contrast MR imaging. Image analysis was done by 2 independent and blinded readers for arterial phase hyperenhancement, washout appearance, enhancing capsule appearance, and size. Hepatic lesions were classified into 7 groups according to LI-RADS v2018.

RESULTS

There was excellent interobserver agreement of both reviewers for LR version 4 (κ = 0.887, P = 0.001) with 90.76% agreement. There was excellent interobserver agreement for nonrim arterial phase hyperenhancement (κ = 0.948; 95% confidence interval [CI], 0.89-0.99; P = 0.001), washout appearance (κ = 0.949; 95% CI, 0.89-1.0; P = 0.001); and enhancing capsule (κ = 0.848; 95% CI, 0.73-0.97; P = 0.001) and excellent reliability of size (interclass correlation, 0.99; P = 0.001). There was excellent interobserver agreement for LR-1 (κ = 1.00, P = 0.001), LR-2 (κ = 0.94, P = 0.001), LR-5 (κ = 0.839, P = 0.001), LR-M (κ = 1.00, P = 0.001), and LR-TIV (κ = 1.00; 95% CI, 1.0-1.0; P = 0.001), and good agreement for LR-3 (κ = 0.61, P = 0.001) and LR-4 (κ = 0.61, P = 0.001).

CONCLUSION

MR imaging of LI-RADS v2018 is a reliable imaging modality and reporting system that may be used for standard interpretation of hepatic focal lesions.

Prevalence and clinical significance of discordant LI-RADS® observations on multiphase contrast-enhanced MRI in patients with cirrhosis

PURPOSE

To determine the prevalence and clinical significance of discordant LI-RADS^® (Liver Imaging Reporting and Data System) liver observations on multiphase contrast-enhanced (MCE) magnetic resonance imaging (MRI) in patients with cirrhosis.

METHODS

This cross-sectional study included 93 cirrhosis patients who underwent 1.5 or 3 T MCE MRI for evaluation of hepatocellular carcinoma (HCC). Two abdominal radiologists independently reviewed T1-, T2-, diffusion-weighted unenhanced images as well as MCE T1-weighted fat-suppressed images and reported liver observations using LI-RADS^®. Concordance were recorded for detection (co-detected by both radiologists or not), size category (< 10; 10-19; ≥ 20 mm), and LI-RADS^® category assignment as reportable (LR-3/4/5/M) and actionable (LR-4/5/M). The overall concordance (i.e., concordant in detection, size, and LR-category) was calculated with 95% confidence interval [CI], and separately for detection, size, and LR-category. Clinical significance of discordance was assessed as impact on follow-up imaging, referral for biopsy, liver transplant eligibility, or treatment modality.

RESULTS

Reportable and actionable observations were overall concordant between two radiologists only in 32.3% [24.6, 41.0] and 40.1% [29.5, 51.5] of cases, respectively. Poor overall concordance was related to detection concordance of 52.0% [44.3, 59.5] and 62.5% [52.3, 71.8], as well as LR-category concordance of 73.7% [61.6, 83.1] and 70.9% [57.3, 81.6], for reportable and actionable observations, respectively. Discordant LI-RADS^® observations would have impacted clinical management in 30 subjects (43.5%), most (66.7%) of whom were due to discordant detection.

CONCLUSION

Discordant MRI LI-RADS^® observations are common in patients with cirrhosis and may have potential implications for patient management.

Subtraction Images From Portal Venous Phase Gadoxetic Acid-Enhanced MRI for Observing Washout and Enhancing Capsule Features in LI-RADS Version 2018

OBJECTIVE. The purpose of this study is to validate the use of subtraction images derived from gadoxetic acid-enhanced MRI for observation of washout and enhancing capsule in the diagnosis of hepatocellular carcinoma (HCC). MATERIALS AND METHODS. For 120 histologically verified HCCs in 115 high-risk patients, the presence of washout and enhancing capsule in the portal venous phase (PVP) on conventional MR images with and without corresponding subtraction images was determined by two independent observers. The incremental value of subtraction imaging in upgrading the categories outlined in the Liver Imaging Reporting and Data System (LI-RADS) version 2018 for the diagnosis of HCC was analyzed for different subgroups of patients classified on the basis of lesion size (< 10 mm, 10-19 mm, ≥ 20 mm), unenhanced T1-weighted signal intensity, and arterial phase hyperenhancement (APHE) of the lesions. RESULTS. When conventional PVP images were compared with the combination of conventional and subtraction PVP images, only T1-weighted isointensity or hyperintensity significantly increased the detection of washout (eight vs 15 of 18 lesions; p = 0.0233). Detection of enhancing capsule was significantly increased (p < 0.05) regardless of various factors, except for a lesion size of less than 2 cm (five vs nine of 36 lesions; p = 0.1336). Two lesions (one LI-RADS category 3 lesion and one LI-RADS category 4 lesion) with APHE were upgraded to category 5 on the basis of a review of PVP subtraction images. CONCLUSION. PVP subtraction imaging during gadoxetic acid-enhanced MRI can upgrade LI-RADS categories for the diagnosis of HCC because of its superior ability in depicting enhancing capsule and the incremental benefit of showing washout.

How high is the inter-observer reproducibility in the LIRADS reporting system?

Purpose

To investigate the reproducibility of LIRADS v2014 and contribute to its widespread use in clinical practice.

Material and methods

This retrospective, single-centre study was conducted between January 2010 and October 2015. A total of 132 patients who had dynamic magnetic resonance imaging (MRI)/computed tomography (CT) images in the Picture Archiving and Communication Systems (PACS) with liver nodule were included in the study, 37 of whom had histopathology results. Five radiologists who participated in the study, interpreted liver nodules independently on different PACS stations according to the LIRADS reporting system and its main parameters.

Results

We determined that level of inter-observer agreement in the LR-1, LR-5, and LR-5V categories was higher than in the LR-2, LR-3, and LR-4 categories (κ = 0.522, 0.442, and 0.600 in the LR-1, LR-5, and LR-5V categories, respectively; κ = 0.082, 0.298, and 0.143 in the LR-2, LR-3, and LR-4 categories, respectively). The parameter that we observed to have the highest level of inter-observer agreement was venous thrombus (κ = 0.600).

Conclusions

Our study showed that LIRADS achieves an acceptable inter-observer reproducibility in terms of clinical practice although it is insufficient at intermediate risk levels. We think that the prevalence of its use will be further increased with training related to the subject and the assignment of numerical values that express the probability of malignancy for each category and including the ancillary features in the algorithm according to clearer rules.

A Semi-Automated Usability Evaluation Framework for Interactive Image Segmentation Systems

For complex segmentation tasks, the achievable accuracy of fully automated systems is inherently limited. Specifically, when a precise segmentation result is desired for a small amount of given data sets, semi-automatic methods exhibit a clear benefit for the user. The optimization of human computer interaction (HCI) is an essential part of interactive image segmentation. Nevertheless, publications introducing novel interactive segmentation systems (ISS) often lack an objective comparison of HCI aspects. It is demonstrated that even when the underlying segmentation algorithm is the same throughout interactive prototypes, their user experience may vary substantially. As a result, users prefer simple interfaces as well as a considerable degree of freedom to control each iterative step of the segmentation. In this article, an objective method for the comparison of ISS is proposed, based on extensive user studies. A summative qualitative content analysis is conducted via abstraction of visual and verbal feedback given by the participants. A direct assessment of the segmentation system is executed by the users via the system usability scale (SUS) and AttrakDiff-2 questionnaires. Furthermore, an approximation of the findings regarding usability aspects in those studies is introduced, conducted solely from the system-measurable user actions during their usage of interactive segmentation prototypes. The prediction of all questionnaire results has an average relative error of 8.9%, which is close to the expected precision of the questionnaire results themselves. This automated evaluation scheme may significantly reduce the resources necessary to investigate each variation of a prototype's user interface (UI) features and segmentation methodologies.

Optimal lexicon of gadoxetic acid-enhanced magnetic resonance imaging for the diagnosis of hepatocellular carcinoma modified from LI-RADS

PURPOSE

To define the optimal lexicon of major imaging findings on gadoxetic acid-enhanced MRIs to diagnose HCC to improve diagnostic performance of the LI-RADS.

METHODS

Two hundred forty-one hepatic lesions (149 HCC, six other malignancies, 86 benign lesions) in 177 treatment-naïve patients at risk of HCC who underwent gadoxetic acid-MRIs from January 2013 to December 2015 were retrospectively reviewed using either histopathological or follow-up imaging findings as a standard reference. Two board-certified radiologists independently evaluated the imaging features and categorized the nodules based on the original and the following modified definitions in LI-RADS: (1) washout appearance in the portal venous phase (PVP) only versus that in the PVP or transitional phase, and (2) enhancing capsule only versus enhancing or non-enhancing capsule. Diagnostic performance and inter-observer agreement of LR-5 were assessed and compared between the algorithms using generalized estimation equation.

RESULTS

The sensitivity [79.2% (95% confidence interval 71.9, 85.0)] and accuracy [84.6% (79.5, 88.7)] of LR-5 were significantly higher for modified lexicon compared with original LI-RADS [60.4% (52.3, 67.9) and 73.9% (67.9, 79.0); P < 0.001 in all cases]. There was no significant difference in specificity [93.5% (86.2, 97.0) and 95.7% (89.0, 98.4); P = 0.153]. Subgroups of lesions < or ≥ 2 cm showed similar tendencies. Inter-observer agreement for capsule appearance was fair to moderate, whereas that for other imaging findings was good to excellent.

CONCLUSIONS

Compared to original LI-RADS, LI-RADS with modified lexicon showed higher sensitivity for the diagnosis of HCC using gadoxetic acid-MRI, with similar specificity.

Magnetic resonance enterography features of small bowel Crohn's disease activity: an inter-rater reliability study of small bowel active inflammation in clinical practice setting

OBJECTIVE

The aim of this study is to determine the interrater agreement in a clinical practice environment for the most commonly used magnetic resonance enterography (MRE) features of Crohn's disease (CD).

METHODS

CD patients with MRE's before and after treatment were retrospectively identified using search queries over a 7-year period (May 2017-September 2017). MRE features of CD comprising components of multiple CD scoring indices were scored by radiologists in the same segment of bowel. Agreement for nominal categorical and continuous variables was assessed using a κ and interclass correlation coefficients, respectively.

RESULTS

80 scans comprised the study population. Moderate interrater agreement was seen in both the pre- and post-treatment MRE's for presence of diffusion restriction (к = 0.43, 0.48; pre- and post-treatment), stricturing disease (к = 0.51, 0.52), overall degree of severity (к = 0.49, 0.59). Substantial agreement was seen in pre- and post-treatment scans for length of involvement (interclass correlation coefficient = 0.67, 0.61). The presence of mucosal ulceration had no agreement (к = -0.07, -0.042).

CONCLUSION

Many MRE features of active CD comprising the major CD scoring indices are reproducible when interpreted by non-CD focused abdominal radiologists. However, the presence of mucosal ulcerations had no agreement and may need more investigation before including this feature as a driver in therapeutic decision making.

ADVANCES IN KNOWLEDGE

Demonstrates the unreliability of mucosal ulceration by non-CD focused abdominal radiologists, targeting a potential area for future education. Key Points The majority of MRE findings incorporated in to many CD scoring indices have fair to moderate inter-rater agreement even when read by non-MRE expert radiologists. Substantial agreement was seen in the length of involved bowel, but this feature is only incorporated in to one of the CD scoring indices. Presence of mucosal ulcerations had no interrater agreement in our study-a feature which is heavily weighted by several CD scoring indices. Research should be focused bridging those features which have poor interrater agreement.

Hepatocellular carcinoma (HCC) versus non-HCC: accuracy and reliability of Liver Imaging Reporting and Data System v2018

PURPOSE

The Liver Imaging Reporting and Data System (LI-RADS) was created to standardize the diagnostic criteria for hepatocellular carcinoma (HCC) and has undergone multiple revisions including a recent update in 2018 (v2018). The primary aim of this study was to determine the diagnostic performance and interrater reliability (IRR) of LI-RADS v2018 for distinguishing HCC from non-HCC primary hepatic malignancy in patients 'at-risk' for HCC. A secondary aim was to assess the impact of changes introduced in the v2018 diagnostic algorithm.

METHODS

This retrospective study combined a 10-year experience of pathologically proven primary liver malignancies from two large liver transplant centers. Two blinded readers independently evaluated each lesion and assigned a LI-RADS diagnostic category, additionally scoring all relevant imaging features. Changes in category based on the reader-provided features and the new v2018 criteria were assessed by a study coordinator.

RESULTS

The final study cohort comprised 105 HCCs and 73 non-HCC primarily liver malignancies. LI-RADS had a high specificity for distinguishing HCC from non-HCC (89% and 90% for reader 1 and reader 2, respectively), and IRR was moderate to substantial for final LI-RADS category and most features. Revision of the LI-RADS v2018 diagnostic algorithm resulted in very few changes [5 (2.8%) and 3 (1.7%) for reader 1 and reader 2, respectively] in overall lesion classification.

CONCLUSION

LI-RADS diagnostic categories and features had moderate to substantial IRR and high specificity for distinguishing HCC from non-HCC primary liver malignancy. Revision of LI-RADS v2018 diagnostic algorithm resulted in reclassification of very few lesions.

Joint Consensus Statement of the Indian National Association for Study of the Liver and Indian Radiological and Imaging Association for the Diagnosis and Imaging of Hepatocellular Carcinoma Incorporating Liver Imaging Reporting and Data System

Hepatocellular carcinoma (HCC) is the 6th most common cancer and the second most common cause of cancer-related mortality worldwide. There are currently no universally accepted practice guidelines for the diagnosis of HCC on imaging owing to the regional differences in epidemiology, target population, diagnostic imaging modalities, and staging and transplant eligibility. Currently available regional and national guidelines include those from the American Association for the Study of Liver Disease (AASLD), the European Association for the Study of the Liver (EASL), the Asian Pacific Association for the Study of the Liver, the Japan Society of Hepatology, the Korean Liver Cancer Study Group, Hong Kong, and the National Comprehensive Cancer Network in the United States. India with its large population and a diverse health infrastructure faces challenges unique to its population in diagnosing HCC. Recently, American Association have introduced a Liver Imaging Reporting and Data System (LIRADS, version 2017, 2018) as an attempt to standardize the acquisition, interpretation, and reporting of liver lesions on imaging and hence improve the coherence between radiologists and clinicians and provide guidance for the management of HCC. The aim of the present consensus was to find a common ground in reporting and interpreting liver lesions pertaining to HCC on imaging keeping LIRADSv2018 in mind.

Validation of Liver Imaging Reporting and Data System 2017 (LI-RADS) Criteria for Imaging Diagnosis of Hepatocellular Carcinoma

BACKGROUND

The Liver Imaging Reporting and Data System (LI-RADS) is being adapted by many clinical practices. To support continuation of its use, LI-RADS (LR) is in need of multicenter validation studies of recent LI-RADS iterations. Furthermore, while both gadoxetate and extracellular agents have been incorporated into LI-RADS, comparison of the diagnostic performance between the two has yet to be determined.

PURPOSE/HYPOTHESIS

To evaluate the rate, diagnostic performance, and interreader reliability (IRR) of LI-RADS 2017 for hepatocellular carcinoma, including LR major and ancillary features, with both gadoxetate and extracellular agent-enhanced MRI against a reference standard of histopathology or imaging follow-up.

STUDY TYPE

Retrospective.

POPULATION

In all, 114 patients with 144 observations were included who met LR 2017 criteria for at risk and had at least one hepatic observation on liver MRI performed with either gadoxetate (n = 52) or an extracellular agent (n = 92) between 2010-2016, with histopathology (n = 103) or follow-up imaging (n = 41).

FIELD STRENGTH/SEQUENCE

1.5 and 3.0T/T₁ -T₂ WI, diffusion-weighted imaging.

ASSESSMENT

Three radiologists independently assessed major/ancillary features and assigned overall LI-RADS category for every observation.

STATISTICAL TESTS

Diagnostic performance of LR5/TIV+LR5 for identifying hepatocellular carcinoma (HCC) was compared between contrast agents with a generalized estimating equation. Weighted kappa was performed for interrater reliability.

RESULTS

The frequency of HCCs among LR1, LR2, LR3, L4, LR5, LRTIV+LR5, and LRM observations were: 0% (all readers), 0-12.5%, 11.4-26.9%, 50-76%, 83.0-95.1%, 83.3-100.0%, and 45.0-65.0%, respectively. Sensitivity of LR5/LRTIV+LR5 for HCC was 59.7-71.4% and specificity 85.0-96.8%. LI-RADS specificity and positive predictive value for observations imaged with gadoxetate was higher than extracellular agent for the most inexperienced reader (R3) (P = 0.009-0.034). IRR for LI-RADS categorization was substantial (k = 0.661).

DATA CONCLUSION

Increasing numerical LI-RADS 2017 categories demonstrate a greater percentage of HCCs. LR5/TIV+LR5 demonstrates excellent specificity and fair sensitivity for HCC. MRI with gadoxetate in liver transplant candidates may be beneficial for less experienced readers, although further large-scale prospective studies are needed.

LEVEL OF EVIDENCE

4 Technical Efficacy: Stage 3 J. Magn. Reson. Imaging 2019;49:e205-e215.

Liver Imaging Reporting and Data System (LI-RADS) Version 2018: Imaging of Hepatocellular Carcinoma in At-Risk Patients

The Liver Imaging Reporting and Data System (LI-RADS) is composed of four individual algorithms intended to standardize the lexicon, as well as reporting and care, in patients with or at risk for hepatocellular carcinoma in the context of surveillance with US; diagnosis with CT, MRI, or contrast material-enhanced US; and assessment of treatment response with CT or MRI. This report provides a broad overview of LI-RADS, including its historic development, relationship to other imaging guidelines, composition, aims, and future directions. In addition, readers will understand the motivation for and key components of the 2018 update.

Pitfalls and problems to be solved in the diagnostic CT/MRI Liver Imaging Reporting and Data System (LI-RADS)

The 2017 Core of the computed tomography (CT)/magnetic resonance imaging (MRI) Liver Imaging Reporting and Data System (LI-RADS) provides clear definitions and concise explanations of the CT/MRI diagnostic algorithm. Nevertheless, there remain some practical and controversial issues that radiologists should be aware of when using the system. This article discusses pitfalls and problems which may be encountered when the version 2017 diagnostic algorithm is used for CT and MRI. The pitfalls include challenges in applying major features and assigning the LR-M category, as well as categorisation discrepancy between CT and MRI. The problems include imprecision of category codes, application of ancillary features, and regional practice variations in hepatocellular carcinoma (HCC) diagnosis. Potential solutions are presented along with these pitfalls and problems. KEY POINTS: • Although the diagnostic algorithm provides clear and detailed explanations, major feature evaluation can be subject to pitfalls and differentiation of HCC and non-HCC malignancy remains challenging. • Ancillary features are optional and equally weighted. However, features such as hepatobiliary phase hypointensity and restricted diffusion have greater impact on HCC diagnosis than other ancillary features and may merit greater emphasis or weighting. • LI-RADS was initially developed from a Western paradigm, which may limit its applicability in the East due to regional practice variations. In Eastern Asia, high sensitivity is prioritised over near-perfect specificity for HCC diagnosis in order to detect tumours at early stages.

Differentiation of Hepatocellular Carcinoma from Other Hepatic Malignancies in Patients at Risk: Diagnostic Performance of the Liver Imaging Reporting and Data System Version 2014

Purpose To evaluate the diagnostic performance and interrater reliability of the Liver Imaging Reporting and Data System (LI-RADS) version 2014 in differentiating hepatocellular carcinoma (HCC) from non-HCC malignancy in a population of patients at risk for HCC. Materials and Methods This retrospective HIPAA-compliant institutional review board-approved study was exempt from informed consent. A total of 178 pathology-proven malignant liver masses were identified in 178 patients at risk for HCC but without established extrahepatic malignancy from August 2012 through August 2015. Two readers blinded to pathology findings and clinical follow-up data independently evaluated a liver protocol magnetic resonance or computed tomography study for each lesion and assigned LI-RADS categories, scoring all major and most ancillary features. Statistical analyses included the independent samples t test, x² test, Fisher exact test, and Cohen k. Results This study included 136 HCCs and 42 non-HCC malignancies. Specificity and positive predictive value of an HCC imaging diagnosis (LR-5 or LR-5V) were 69.0% and 90.5%, respectively, for reader 1 (R1) and 88.3% and 95.5%, respectively, for reader 2 (R2). Tumor in vein was a common finding in patients with non-HCC malignancies (R1, 10 of 42 [23.8%]; R2, five of 42 [11.9%]). Exclusion of the LR-5V pathway improved specificity and positive predictive value for HCC to 83.3% and 92.9%, respectively, for R1 (six fewer false-positive findings) and 92.3% and 96.4%, respectively, for R2 (one fewer false-positive finding). Among masses with arterial phase hyperenhancement, the rim pattern was more common among non-HCC malignancies than among HCCs for both readers (R1: 24 of 36 [66.7%] vs 13 of 124, [10.5%], P < .001; R2: 27 of 35 [77.1%] vs 21 of 123 [17.1%], P < .001) (k = 0.76). Exclusion of rim arterial phase hyperenhancement as a means of satisfying LR-5 criteria also improved specificity and positive predictive value for HCC (R1, two fewer false-positive findings). Conclusion Modification of the algorithmic role of tumor in vein and rim arterial phase hyperenhancement improves the diagnostic performance of LI-RADS version 2014 in differentiating HCC from non-HCC malignancy. ^© RSNA, 2017 Online supplemental material is available for this article.

Diagnostic efficacy of the Liver Imaging-Reporting and Data System (LI-RADS) with CT imaging in categorising small nodules (10-20 mm) detected in the cirrhotic liver at screening ultrasound

AIM

To estimate the diagnostic accuracy of the Liver Imaging-Reporting and Data System (LI-RADS) with computed tomography (CT) for diagnosing hepatic nodules (10-20 mm) detected in cirrhotic livers.

MATERIALS AND METHODS

Fifty-five patients with liver cirrhosis and a solitary nodule (10-20 mm in diameter) detected via ultrasound surveillance, underwent hepatic CT and fine-needle biopsy. All the CT images were analysed and the lesions were categorised into five categories according to the LI-RADS.

RESULTS

Final diagnoses of the 55 nodules were as follows: 34 hepatocellular carcinomas (HCCs), one intrahepatic cholangiocarcinomas, one adrenocortical carcinoma metastasis, and 19 benign lesions. None (0%) of four LI-RADS category 1 lesions, two (22%) of nine category 2 lesions, seven (50%) of 14 category 3 lesions, two (67%) of three category 4 lesions, 22 (96%) of 23 category 5 lesions and one (50%) of two lesions classified as other malignancies was HCC. One category 5 lesion was adrenocortical carcinoma metastasis and one of two lesions categorised as other malignancies was intrahepatic cholangiocarcinoma. In patients with nodules detected at surveillance ultrasound, the best threshold for confident HCC diagnosis was more than LI-RADS category 3. The use of this threshold produced a sensitivity and specificity of 72.7% and 90%, respectively. So combining LI-RADS 4 and 5 categories for confident HCC diagnosis would improve accuracy and sensitivity with no significant impairment of specificity or positive predictive value.

CONCLUSION

LIRADS with CT provides a strong validity for the diagnosis of small hepatic nodules, and is very useful to improve the accuracy of CT reports.

Major and ancillary magnetic resonance features of LI-RADS to assess HCC: an overview and update

Liver Imaging Reporting and Data System (LI-RADS) is a system for interpreting and reporting of imaging features on multidetector computed tomography (MDCT) and magnetic resonance (MR) studies in patients at risk for hepatocellular carcinoma (HCC). American College of Radiology (ACR) sustained the spread of LI-RADS to homogenizing the interpreting and reporting data of HCC patients. Diagnosis of HCC is due to the presence of major imaging features. Major features are imaging data used to categorize LI-RADS-3, LI-RADS-4, and LI-RADS-5 and include arterial-phase hyperenhancement, tumor diameter, washout appearance, capsule appearance and threshold growth. Ancillary are features that can be used to modify the LI-RADS classification. Ancillary features supporting malignancy (diffusion restriction, moderate T2 hyperintensity, T1 hypointensity on hapatospecifc phase) can be used to upgrade category by one or more categories, but not beyond LI-RADS-4. Our purpose is reporting an overview and update of major and ancillary MR imaging features in assessment of HCC.

Critical analysis of the major and ancillary imaging features of LI-RADS on 127 proven HCCs evaluated with functional and morphological MRI: Lights and shadows

PURPOSE

To report a critical analysis of major and ancillary MR imaging features in assessment of HCC.

METHODS

Retrospectively we evaluated 70 cirrhotic patients with 173 nodules, which were subjected to MR study at 0 time (MR0), after 3 (MR3) and 6 months (MR6) using two different contrast media. EOB-GD-DTPA was injected at MR0 and MR6, while Gd-BT-DO3A at MR3. Three expert hepatic radiologists reviewed all images, recording, according to LI-RADS, the size, the presence and quality of arterial-phase hyperenhancement, washout and capsule appearance, threshold growth. Additionally, we recorded signal intensity (SI) on T2-W images, on DWI, on apparent diffusion coefficient (ADC) maps and SI on T1-W images of EOB-GD-BPTA hepatospecific phase. Median value of ADC and of Intravoxel incoherent motion related parameters were assessed.

RESULTS

127 HCCs and 24 dysplastic nodules were assessed. Hypervascular on arterial phase was found in 84 HCCs, washout appearance in 124, capsule appearance in 111, hypointensity on hepatospecific phase in 127, hyperintensity on T2-W sequences and restricted diffusion in 107. Hyper vascular on arterial phase was found in 17 dysplastic nodules, wash-out appearance in 2, hypointensity on hepatospecific phase in 7 while no dysplastic nodules showed capsule appearance, hyperintensity on T2-W and restricted diffusion. Highest accuracy was obtained by washout appearance and hypointense signal on hepatospecific phase (97% and 95%).

CONCLUSIONS

Hypointensity on hepatospecific phase and washout appearance are the most relevant diagnostic sign for differentiating low-risk from high-risk HCC nodules. The capsule appearance, T2-W hyperintensity and restricted diffusion have high positive predictive value.