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.

Performance of LI-RADS category 5 vs combined categories 4 and 5: a systemic review and meta-analysis.

OBJECTIVE: Computed tomography (CT)/magnetic resonance imaging (MRI) Liver Imaging Reporting and Data System (LI-RADS, LR) category 5 has high specificity and modest sensitivity for diagnosis of hepatocellular carcinoma (HCC). The purpose of this study was to compare the diagnostic performance of LR-5 vs combined LR-4 and LR-5 (LR-4/5) for HCC diagnosis. METHODS: MEDLINE and EMBASE databases through January 03, 2023 were searched for studies reporting the performance of LR-5 and combined LR-4/5 for HCC diagnosis, using CT/MRI LI-RADS version 2014, 2017, or 2018. A bivariate random-effects model was used to calculate the pooled per-observation diagnostic performance. Subgroup analysis was performed based on imaging modalities and type of MRI contrast material. RESULTS: Sixty-nine studies (15,108 observations, 9928 (65.7%) HCCs) were included. Compared to LR-5, combined LR-4/5 showed significantly higher pooled sensitivity (83.0% (95% CI [80.3-85.8%]) vs 65.7% (95% CI [62.4-69.1%]); p < 0.001), lower pooled specificity (75.0% (95% CI [70.5-79.6%]) vs 91.7% (95% CI [90.2-93.1%]); p < 0.001), lower pooled positive likelihood ratio (3.60 (95% CI [3.06-4.23]) vs 6.18 (95% CI [5.35-7.14]); p < 0.001), and lower pooled negative likelihood ratio (0.22 (95% CI [0.19-0.25]) vs 0.38 (95% CI [0.35-0.41]) vs; p < 0.001). Similar results were seen in all subgroups. CONCLUSIONS: Our meta-analysis showed that combining LR-4 and LR-5 would increase sensitivity but decrease specificity, positive likelihood ratio, and negative likelihood ratio. These findings may inform management guidelines and individualized management. CLINICAL RELEVANCE STATEMENT: This meta-analysis estimated the magnitude of changes in the sensitivity and specificity of imaging criteria when LI-RADS categories 4 and 5 were combined; these findings can inform management guidelines and individualized management. KEY POINTS: There is no single worldwide reporting system for liver imaging, partly due to regional needs. Combining LI-RADS categories 4 and 5 increased sensitivity and decreased specificity and positive and negative likelihood ratios. Changes in the sensitivity and specificity of imaging criteria can inform management guidelines and individualized management.

Radiomics model versus 2017 revised international consensus guidelines for predicting malignant intraductal papillary mucinous neoplasms.

OBJECTIVES: To evaluate a CT-based radiomics model for identifying malignant pancreatic intraductal papillary mucinous neoplasms (IPMNs) and compare its performance with the 2017 international consensus guidelines (ICGs). MATERIALS AND METHODS: We retrospectively included 194 consecutive patients who underwent surgical resection of pancreatic IPMNs between January 2008 and December 2020. Surgical histopathology was the reference standard for diagnosing malignancy. Using radiomics features from preoperative contrast-enhanced CT, a radiomics model was built with the least absolute shrinkage and selection operator by a five-fold cross-validation. CT and MR images were independently reviewed based on the 2017 ICGs by two abdominal radiologists, and the performances of the 2017 ICGs and radiomics model were compared. The areas under the curve (AUCs) were compared using the DeLong method. RESULTS: A total of 194 patients with pancreatic IPMNs (benign, 83 [43%]; malignant, 111 [57%]) were chronologically divided into training (n = 141; age, 65 ± 8.6 years; 88 males) and validation sets (n = 53; age, 66 ± 9.7 years; 31 males). There was no statistically significant difference in the diagnostic performance of the 2017 ICGs between CT and MRI (AUC, 0.71 vs. 0.71; p = 0.93) with excellent intermodality agreement (k = 0.86). In the validation set, the CT radiomics model had higher AUC (0.85 vs. 0.71; p = 0.038), specificity (84.6% vs. 61.5%; p = 0.041), and positive predictive value (84.0% vs. 66.7%; p = 0.044) than the 2017 ICGs. CONCLUSION: The CT radiomics model exhibited better diagnostic performance than the 2017 ICGs in classifying malignant IPMNs. CLINICAL RELEVANCE STATEMENT: Compared with the radiologists' evaluation based on the 2017 international consensus guidelines, the CT radiomics model exhibited better diagnostic performance in classifying malignant intraductal papillary mucinous neoplasms. KEY POINTS: • There is a paucity of comparisons between the 2017 international consensus guidelines (ICGs) and radiomics models for malignant intraductal papillary mucinous neoplasms (IPMNs). • The CT radiomics model developed in this study exhibited better diagnostic performance than the 2017 ICGs in classifying malignant IPMNs. • The radiomics model may serve as a valuable complementary tool to the 2017 ICGs, potentially allowing a more quantitative assessment of IPMNs.

Liver Imaging Reporting and Data System version 2018 category 5 for diagnosing hepatocellular carcinoma: an updated meta-analysis.

OBJECTIVE: We performed an updated meta-analysis to determine the diagnostic performance of Liver Imaging Reporting and Data System (LI-RADS, LR) 5 category for hepatocellular carcinoma (HCC) using LI-RADS version 2018 (v2018), and to evaluate differences by imaging modalities and type of MRI contrast material. METHODS: The MEDLINE and Embase databases were searched for studies reporting the performance of LR-5 using v2018 for diagnosing HCC. A bivariate random-effects model was used to calculate the pooled per-observation sensitivity and specificity. Subgroup analysis was performed based on imaging modalities and type of MRI contrast material. RESULTS: Forty-eight studies qualified for the meta-analysis, comprising 9031 patients, 10,547 observations, and 7216 HCCs. The pooled per-observation sensitivity and specificity of LR-5 for diagnosing HCC were 66% (95% CI, 61-70%) and 91% (95% CI, 89-93%), respectively. In the subgroup analysis, MRI with extracellular agent (ECA-MRI) showed significantly higher pooled sensitivity (77% [95% CI, 70-82%]) than CT (66% [95% CI, 58-73%]; p = 0.023) or MRI with gadoxetate (Gx-MRI) (65% [95% CI, 60-70%]; p = 0.001), but there was no significant difference between ECA-MRI and MRI with gadobenate (gadobenate-MRI) (73% [95% CI, 61-82%]; p = 0.495). Pooled specificities were 88% (95% CI, 80-93%) for CT, 92% (95% CI, 86-95%) for ECA-MRI, 93% (95% CI, 91-95%) for Gx-MRI, and 91% (95% CI, 84-95%) for gadobenate-MRI without significant differences (p = 0.084-0.803). CONCLUSIONS: LI-RADS v2018 LR-5 provides high specificity for HCC diagnosis regardless of modality or contrast material, while ECA-MRI showed higher sensitivity than CT or Gx-MRI. CLINICAL RELEVANCE STATEMENT: Refinement of the criteria for improving sensitivity while maintaining high specificity of LR-5 for HCC diagnosis may be an essential future direction. KEY POINTS: • The pooled per-observation sensitivity and specificity of LR-5 for diagnosing HCC using LI-RADSv2018 were 66% and 91%, respectively. • ECA-MRI showed higher sensitivity than CT (77% vs 66%, p = 0.023) or Gx-MRI (77% vs 65%, p = 0.001). • LI-RADS v2018 LR-5 provides high specificity (88-93%) for HCC diagnosis regardless of modality or contrast material type.

Prognostic significance of MRI features in patients with solitary large hepatocellular carcinoma following surgical resection.

OBJECTIVE: To assess the prognostic impact of preoperative MRI features on outcomes for single large hepatocellular carcinoma (HCC) (≥ 8 cm) after surgical resection. MATERIAL AND METHODS: This retrospective study included 151 patients (mean age: 59.2 years; 126 men) with a single large HCC who underwent gadoxetic acid-enhanced MRI and surgical resection between 2008 and 2020. Clinical variables, including tumor markers and MRI features (tumor size, tumor margin, and the proportion of hypovascular component on hepatic arterial phase (AP) (≥ 50% vs. < 50% tumor volume) were evaluated. Cox proportional hazards model analyzed overall survival (OS), recurrence-free survival (RFS), and associated factors. RESULTS: Among 151 HCCs, 37.8% and 62.2% HCCs were classified as ≥ 50% and < 50% AP hypovascular groups, respectively. The 5- and 10-year OS and RFS rates in all patients were 62.0%, 52.6% and 41.4%, 38.5%, respectively. Multivariable analysis revealed that ≥ 50% AP hypovascular group (hazard ratio [HR] 1.7, p = 0.048), tumor size (HR 1.1, p = 0.006), and alpha-fetoprotein ≥ 400 ng/mL (HR 2.6, p = 0.001) correlated with poorer OS. ≥ 50% AP hypovascular group (HR 1.9, p = 0.003), tumor size (HR 1.1, p = 0.023), and non-smooth tumor margin (HR 2.1, p = 0.009) were linked to poorer RFS. One-year RFS rates were lower in the ≥ 50% AP hypovascular group than in the < 50% AP hypovascular group (47.4% vs 66.9%, p = 0.019). CONCLUSION: MRI with ≥ 50% AP hypovascular component and larger tumor size were significant factors associated with poorer OS and RFS after resection of single large HCC (≥ 8 cm). These patients require careful multidisciplinary management to determine optimal treatment strategies. CLINICAL RELEVANCE STATEMENT: Preoperative MRI showing a ≥ 50% arterial phase hypovascular component and larger tumor size can predict worse outcomes after resection of single large hepatocellular carcinomas (≥ 8 cm), underscoring the need for tailored, multidisciplinary treatment strategies. KEY POINTS: MRI features offer insights into the postoperative prognosis for large hepatocellular carcinoma. Hypovascular component on arterial phase ≥ 50% and tumor size predicted poorer overall survival and recurrence-free survival. These findings can assist in prioritizing aggressive and multidisciplinary approaches for patients at risk for poor outcomes.

Percentages of Hepatocellular Carcinoma in LI-RADS Categories with CT and MRI: A Systematic Review and Meta-Analysis.

Background The Liver Imaging Reporting and Data System (LI-RADS) CT and MRI algorithm applies equally to CT, MRI with extracellular contrast agents (ECA-MRI), and MRI with gadoxetate (Gx-MRI). Purpose To estimate pooled percentages of hepatocellular carcinoma (HCC) and overall malignancy for each LI-RADS category with CT and MRI. Materials and Methods MEDLINE and EMBASE databases were searched for research articles (January 2014-April 2021) reporting the percentages of observations in each LI-RADS category with use of versions 2014, 2017, or 2018. Study design, population characteristics, imaging modality, reference standard, and numbers of HCC and non-HCC malignancies in each category were recorded. A random-effects model evaluated the pooled percentage of HCC and overall malignancy for each category. Results There were 49 studies with 9620 patients and a total 11 562 observations, comprising 7921 HCCs, 1132 non-HCC malignancies, and 2509 benign entities. No HCC or non-HCC malignancies were reported with any modality in the LR-1 category. The pooled percentages of HCC for CT, ECA-MRI, and Gx-MRI, respectively, were 10%, 6%, and 1% for LR-2 (P = .16); 48%, 31%, and 38% for LR-3 (P = .42); 76%, 64%, and 77% for LR-4 (P = .62); 96%, 95%, and 96% for LR-5 (P = .76); 88%, 76%, and 78% for LR-5V or LR-TIV (tumor in vein) (P = .42); and 20%, 30%, and 35% for LR-M (P = .32). Most LR-M (93%-100%) and LR-5V or LR-TIV (99%-100%) observations were malignant, regardless of modality. Conclusion There was no difference in percentages of hepatocellular carcinoma and overall malignancy between CT, MRI with extracellular contrast agents, and MRI with gadoxetate for any Liver Imaging Reporting and Data System categories. © RSNA, 2023 Supplemental material is available for this article See also the editorial by Ronot in this issue.

Diagnostic Performance of the 2018 EASL vs. LI-RADS for Hepatocellular Carcinoma Using CT and MRI: A Systematic Review and Meta-Analysis of Comparative Studies.

BACKGROUND: Hepatocellular carcinoma (HCC) can be diagnosed without pathologic confirmation in high-risk patients. Therefore, it is necessary to compare current imaging criteria for noninvasive-diagnosis of HCC. PURPOSE: To systematically compare performance of 2018 European Association for the Study of the Liver (EASL) criteria and Liver Imaging Reporting and Data System (LI-RADS) for noninvasive-diagnosis of HCC. STUDY TYPE: Systematic review and meta-analysis. SUBJECTS: Eight studies with 2232 observations, including 1617 HCCs. FIELD STRENGTH/SEQUENCE: 1.5 T, 3.0 T/T2-weighted, unenhanced T1-weighted in-/opposed-phases, multiphase T1-weighted imaging. ASSESSMENT: Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, two reviewers independently reviewed and extracted data, including patient characteristics, index test, reference standard and outcomes, from studies intraindividually comparing the sensitivities and specificities of 2018 EASL-criteria and LR-5 of LI-RADS for HCC. Risk of bias and concerns regarding applicability were evaluated using QUADAS-2 tool. Subgroup analysis was performed based on observation size (≥20 mm, 10-19 mm). STATISTICAL TESTS: Bivariate random-effects model to calculate pooled per-observation sensitivity and specificity of both imaging criteria, and pooled estimates of intraindividual paired data were compared considering the correlation. Forest and linked-receiver-operating-characteristic plots were drawn, and study heterogeneity was assessed using Q-test and Higgins-index. Publication bias was evaluated by Egger's test. A P-value <0.05 was considered statistically significant, except for heterogeneity (P < 0.10). RESULTS: The sensitivity for HCC did not differ significantly between the imaging-based diagnosis using EASL-criteria (61%; 95% CI, 50%-73%) and LR-5 (64%; 95% CI, 53%-76%; P = 0.165). The specificities were also not significantly different between EASL-criteria (92%; 95% CI, 89%-94%) and LR-5 (94%; 95% CI, 91%-96%; P = 0.257). In subgroup analysis, no statistically significant differences were identified in the pooled performances between the two criteria for observations ≥20 mm (sensitivity P = 0.065; specificity P = 0.343) or 10-19 mm (sensitivity P > 0.999; specificity P = 0.851). There was no publication bias for EASL (P = 0.396) and LI-RADS (P = 0.526). DATA CONCLUSION: In the present meta-analysis of paired comparisons, the pooled sensitivities and specificities were not significantly different between 2018 EASL-criteria and LR-5 of LI-RADS for noninvasive-diagnosis of HCC. EVIDENCE LEVEL: 3. TECHNICAL EFFICACY: Stage 2.

LI-RADS Category on MRI Is Associated With Recurrence of Intrahepatic Cholangiocarcinoma After Surgery: A Multicenter Study.

BACKGROUND: The Liver Imaging Reporting and Data System (LI-RADS) is a comprehensive system for standardizing the terminology and interpretation of liver imaging. The association between the LI-RADS category and tumor recurrence in patients with intrahepatic cholangiocarcinomas (iCCAs) has not yet been evaluated in a multicenter study. PURPOSE: To retrospectively investigate the preoperative clinical and imaging features associated with recurrence-free survival (RFS) after curative resection of iCCAs and to identify the role of the LI-RADS category in at-risk patients. STUDY TYPE: Retrospective, multicenter. SUBJECTS: A total of 113 patients (mean age: 61.1 years; 74 men, 39 women) who underwent preoperative contrast-enhanced MRI and curative surgical resection for a single treatment-naive iCCA between 2008 and 2021. FILED STRENGTH/SEQUENCE: A 3 T dual gradient-echo T1 WI with in- and opposed-phase, turbo spin-echo T2 WI, diffusion-weighted echo-planar images, and three-dimensional gradient-echo T1 WI before and after administration of contrast agent. ASSESSMENT: MR imaging features were evaluated and assigned for each lesion using LI-RADS version 2018. RFS was calculated from the date of surgery to tumor recurrence or the last imaging date without evidence of recurrence. Factors affecting RFS were evaluated using clinical and imaging features. STATISTICAL TESTS: Cox proportional hazards model, Kaplan-Meier method, and log-rank test. A P-value of <0.05 was considered statistically significant. RESULTS: A total of 93 (82.3%) were categorized as LR-M and 20 (17.7%) were categorized as LR-4 or 5. In the multivariable analysis, LR-M category (hazard ratio [HR], 8.035; 95% confidence interval [CI], 1.096-58.931) and a tumor size >3 cm on MRI (HR, 2.690; 95% CI, 1.319-5.487) were independent factors for poor RFS. The 5-year RFS rate was significantly higher in patients with iCCA categorized as LR-4 or 5 than in those categorized as LR-M (94.4% vs. 51.9%, respectively). DATA CONCLUSION: Patients with iCCA categorized as LR-4 or 5 may have a better RFS than those categorized as LR-M. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

MRI Radiomics Model Predicts Pathologic Complete Response of Rectal Cancer Following Chemoradiotherapy.

Background Preoperative assessment of pathologic complete response (pCR) in locally advanced rectal cancer (LARC) after neoadjuvant chemoradiotherapy (nCRT) is increasingly needed for organ preservation, but large-scale validation of an MRI radiomics model remains lacking. Purpose To evaluate radiomics models based on T2-weighted imaging and diffusion-weighted MRI for predicting pCR after nCRT in LARC and compare their performance with visual assessment by radiologists. Materials and Methods This retrospective study included patients with LARC (clinical stage T3 or higher, positive nodal status, or both) who underwent post-nCRT MRI and elective resection between January 2009 and December 2018. Surgical histopathologic analysis was the reference standard for pCR. Radiomic features were extracted from the volume of interest on T2-weighted images and apparent diffusion coefficient (ADC) maps from post-nCRT MRI to generate three models: T2 weighted, ADC, and both T2 weighted and ADC (merged). Radiomics signatures were generated using the least absolute shrinkage and selection operator with tenfold cross-validation. Three experienced radiologists independently rated tumor regression grades at MRI and compared these with the radiomics models' diagnostic outcomes. Areas under the curve (AUCs) of the radiomics models and pooled readers were compared by using the DeLong method. Results Among 898 patients, 189 (21%) achieved pCR. The patients were chronologically divided into training (n = 592; mean age ± standard deviation, 59 years ± 12; 388 men) and test (n = 306; mean age, 59 years ± 12; 190 men) sets. The radiomics signatures of the T2-weighted, ADC, and merged models demonstrated AUCs of 0.82, 0.79, and 0.82, respectively, with no evidence of a difference found between the T2-weighted and merged models (P = .49), while the ADC model performed worse than the merged model (P = .02). The T2-weighted model had higher classification performance (AUC, 0.82 vs 0.74 [P = .009]) and sensitivity (80.0% vs 15.6% [P < .001]), but lower specificity (68.4% vs 98.6% [P < .001]) than the pooled performance of the three radiologists. Conclusion An MRI-based radiomics model showed better classification performance than experienced radiologists for diagnosing pathologic complete response in patients with locally advanced rectal cancer after neoadjuvant chemoradiotherapy. © RSNA, 2022 Online supplemental material is available for this article. See also the editorial by Taylor in this issue.

Diagnostic performance of CT versus MRI Liver Imaging Reporting and Data System category 5 for hepatocellular carcinoma: a systematic review and meta-analysis of comparative studies.

OBJECTIVE: To compare the performance of Liver Imaging Reporting and Data System category 5 (LR-5) for diagnosing HCC between CT and MRI using comparative studies. METHODS: The MEDLINE and EMBASE databases were searched from inception to April 21, 2021, to identify studies that directly compare the diagnostic performance of LR-5 for HCC between CT and MRI. A bivariate random-effects model was fitted to calculate the pooled per-observation sensitivity and specificity of LR-5 of each modality, and compare the pooled estimates of paired data. Subgroup analysis was performed according to the MRI contrast agent. RESULTS: Seven studies with 1145 observations (725 HCCs) were included in the final analysis. The pooled per-observation sensitivity of LR-5 for diagnosing HCC was higher using MRI (61%; 95% confidence interval [CI], 43-76%; I2 = 95%) than CT (48%; 95% CI, 31-65%; I2 = 97%) (p < 0.001). The pooled per-observation specificities of LR-5 did not show statistically significant difference between CT (96%; 95% CI, 92-98%; I2 = 0%) and MRI (93%; 95% CI, 88-96%; I2 = 16%) (p = 0.054). In the subgroup analysis, extracellular contrast agent-enhanced MRI showed significantly higher pooled per-observation sensitivity than gadoxetic acid-enhanced MRI for diagnosing HCC (73% [95% CI, 55-85%] vs. 55% [95% CI, 39-70%]; p = 0.007), without a significant difference in specificity (93% [95% CI, 80-98%] vs. 94% [95% CI, 87-97%]; p = 0.884). CONCLUSIONS: The LR-5 of MRI showed significantly higher pooled per-observation sensitivity than CT for diagnosing HCC. The pooled per-observation specificities of LR-5 were comparable between the two modalities. KEY POINTS: • The pooled sensitivity of LR-5 using MRI was higher than that using CT (61% versus 48%), but the pooled specificities of LR-5 were not significantly different between CT and MRI (96% versus 93%). • Subgroup analysis according to the MRI contrast media showed a significantly higher pooled per-observation sensitivity using ECA-enhanced MRI than with EOB-enhanced MRI (73% versus 55%), and comparable specificities (93% versus 94%). • Although LI-RADS provides a common diagnostic algorithm for CT or MRI, the per-observation performance of LR-5 can be affected by the imaging modality as well as the MRI contrast agent.

MRI-diagnosis of category LR-M observations in the Liver Imaging Reporting and Data System v2018: a systematic review and meta-analysis.

OBJECTIVES: We performed a meta-analysis to determine the probability of hepatocellular carcinoma (HCC) and non-HCC malignancies in Liver Imaging Reporting and Data System (LI-RADS) category M (LR-M) observations and the frequency of defined LR-M imaging features on MRI using LI-RADS v2018. METHODS: We searched the MEDLINE and EMBASE databases to identify studies published from 1 January 2018 to 16 March 2021 reporting the probability of category LR-M in HCC and non-HCC malignancies on MRI. The pooled percentages of HCC and non-HCC malignancies in the LR-M observations were evaluated. Meta-regression analysis was performed to identify factors for study heterogeneity. The frequencies of defined LR-M imaging features were also calculated. Risk of bias and concerns regarding applicability were evaluated using the Quality Assessment of Diagnostic Accuracy Studies-2 tool. RESULTS: We identified 18 studies reporting the diagnostic performance of the LR-M category (3,812 observations in 3,615 patients), with nine studies reporting the frequencies of LR-M imaging features. The pooled percentages of HCC and non-HCC malignancies in the LR-M observations were 29% (95% confidence interval [CI], 21-38%) and 67% (95%CI, 57-77%), respectively. The study type and inclusion of benign lesions were significant factors for study heterogeneity. Of the 10 LR-M imaging features, rim arterial phase hyperenhancement (APHE) showed the highest frequency in non-HCC malignancies (68%; 95%CI, 61-75%). CONCLUSIONS: The LR-M category was commonly used to characterize non-HCC malignancies, but also included 29% of HCC. The frequencies of the different LR-M imaging features were variable, with rim APHE showing the highest frequency in non-HCC malignancies. KEY POINTS: • In the LR-M category using LI-RADS v2018 for MRI, the pooled percentage of malignancies in general was 96%, with 29% HCC and 67% non-HCC malignancies, while the remaining 4% was benign entity. • The study type and inclusion of benign lesions were significant factors contributing to substantial heterogeneity among included studies. • The frequencies of the different LR-M imaging features were variable, with rim APHE showing the highest frequency in non-HCC malignancies.

Incorporation of Ancillary MRI Features Into the LI-RADS Treatment Response Algorithm: Impact on Diagnostic Performance After Locoregional Treatment of Hepatocellular Carcinoma.

BACKGROUND. The LI-RADS treatment response algorithm may lack sufficient sensitivity for viable tumor after locoregional treatment (LRT) for hepatocellular carcinoma (HCC). OBJECTIVE. The purpose of our study was to evaluate the impact of incorporation of ancillary MRI features on the diagnostic performance of the LI-RADS treatment response algorithm after LRT for HCC. METHODS. This retrospective study included 141 patients (114 men, 27 women; median age, 56 years) who underwent gadoxetic acid-enhanced MRI after LRT for HCC between October 2005 and January 2020 and subsequent liver surgery. Two readers assessed lesions for LI-RADS features of viability for ancillary MRI features (transitional phase [TP] hypointensity, hepatobiliary phase [HBP] hypointensity, DWI hyperintensity or low ADC, and mild-to-moderate T2 hyperintensity). Interobserver agreement was assessed before reaching consensus. Significant ancillary features were identified using random forest analysis. The impact of incorporation of significant ancillary features on diagnostic performance for incomplete pathologic necrosis (IPN; pathologically viable tumor > 0 mm) was assessed using McNemar tests. RESULTS. Complete pathologic necrosis (CPN) was observed in 88 of 181 (48.6%) lesions. Interreader agreement was almost perfect for LI-RADS features of viability (κ = 0.92-0.97) and was substantial to almost perfect for ancillary features (κ = 0.73-0.94). Random forest analysis identified TP hypointensity (present in 8.0%, 25.0%, and 75.3% of lesions with CPN, viable tumor < 10 mm, and viable tumor ≥ 10 mm, respectively) and HBP hypointensity (9.2%, 25.0%, and 74.0%, respectively) as significant ancillary features. For detecting IPN, LR-TR (treatment response) Viable or LR-TR Equivocal had higher sensitivity than LR-TR Viable (71.0% vs 57.0%, respectively; p = .001) but had lower specificity (86.4% vs 94.3%, p = .02). However, LR-TR Viable or LR-TR Equivocal and TP hypointensity showed higher sensitivity than LR-TR Viable (64.5% vs 57.0%, p = .02) without a significantly different specificity (90.9% vs 94.3%, p = .25). LR-TR Viable or LR-TR Equivocal and HBP hypointensity also showed higher sensitivity than LR-TR Viable (65.6% vs 57.0%, p = .01) without a significantly different specificity (90.8% vs 94.3%, p = .25). CONCLUSION. TP hypointensity and HBP hypointensity increase the sensitivity of LI-RADS treatment response algorithm for viable tumor without lowering specificity. CLINICAL IMPACT. The two identified ancillary features may improve tumor viability assessment and planning of additional therapies after LRT for HCC.

LI-RADS Major Features on MRI for Diagnosing Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis.

BACKGROUND: The reported diagnostic performance for hepatocellular carcinoma (HCC) of each major imaging feature on MRI using standardized definitions of the Liver Imaging Reporting and Data System (LI-RADS) is variable. It is important to know the actual performance of each LI-RADS major imaging feature for imaging diagnosis of HCC and determine the sources of heterogeneity between the reported results. PURPOSE: To systematically determine the performance of each major imaging feature of LI-RADS for diagnosing HCC using either extracellular contrast agent-enhanced MRI (ECA-MRI) or gadoxetate disodium-enhanced MRI (EOB-MRI). STUDY TYPE: Systematic review and meta-analysis. SUBJECTS: Sixteen original articles with 3542 lesions. FIELD STRENGTH: A 1.5 T and 3.0 T. ASSESSMENT: Data extraction was independently performed by two reviewers who identified and reviewed original articles reporting the diagnostic performance of each LI-RADS major imaging feature-arterial phase hyperenhancement (APHE), observation size, washout appearance, enhancing "capsule," and threshold growth-using MRI. Study characteristics, study population characteristics, MRI characteristics, contrast agent, LI-RADS version, reference standards, and study outcomes were extracted from included studies. Risk of bias and concerns regarding applicability were evaluated using the Quality Assessment of Diagnostic Accuracy Studies-2 tool. STATISTICAL TESTS: Bivariate random-effects models were used to obtain summary estimates of the diagnostic performance of each LI-RADS major imaging feature. Hierarchical summary receiver operating characteristic curves were plotted. Meta-regression analyses were performed to explore potential sources of heterogeneity. RESULTS: The pooled per-observation sensitivities and specificities for diagnosing HCC were 85% (95% confidence interval [CI] = 78%-89%) and 57% (95% CI = 44%-70%) for arterial phase hyperenhancement (APHE), 77% (95% CI = 72%-82%), and 74% (95% CI = 63%-83%) for washout appearance, and 52% (95% CI = 41%-64%) and 90% (95% CI = 85%-94%) for enhancing "capsule," respectively. DATA CONCLUSIONS: Among the LI-RADS major features, the sensitivity was the highest for APHE and the specificity was the highest for enhancing "capsule" in the diagnosis of HCC. EVIDENCE LEVEL: 3 TECHNICAL EFFICACY: Stage 2.

Characteristics and Early Recurrence of Hepatocellular Carcinomas Categorized as LR-M: Comparison with Those Categorized as LR-4 or 5.

BACKGROUND: According to the Liver Imaging Reporting and Data System (LI-RADS), the LI-RADS category M (LR-M), which are probably or definitely malignant but are not specific for hepatocellular carcinomas (HCCs), does not exclude HCCs. A gap in knowledge remains, including their characteristics and recurrence of HCCs categorized as LR-M. PURPOSE: To compare the characteristics of HCCs categorized as LR-M with HCCs categorized as LR-4 or LR-5 (LR-4/5) using the LI-RADS version 2018 and evaluate the relationship of these categories with the risk of early recurrence after curative resections of single HCCs. STUDY TYPE: Retrospective. SUBJECTS: Two hundred and eighty-one patients (mean age, 57 years; 191 men and 90 women) who underwent curative resections for single HCCs and preoperative contrast-enhanced MRI between 2015 and 2017. FIELD STRENGTH/SEQUENCE: 3T Dual gradient-echo T1 WI with in- and opposed-phase, turbo spin-echo T2 WI, diffusion-weighted echo-planar images, and three-dimensional gradient-echo T1 WI before and after administration of contrast agent. ASSESSMENT: MRI features according to the LI-RADS version 2018 were evaluated and LI-RADS category were assigned for each observation. Clinical, imaging, and histopathological features were compared based on LI-RADS categorization. Early recurrence rates (<2 years) and associated factors were also evaluated. STATISTICAL TESTS: Fisher's exact test, two-sample t test after satisfying assumption of normality through Shapiro-Wilk test, Fleiss κ coefficient, Cox proportional hazards regression analysis, Kaplan-Meier method, and log-rank test. RESULTS: Forty-one HCCs (14.6%) were categorized as LR-M and 240 HCCs (85.4%) were categorized as LR-4/5. LR-M HCCs showed poorer differentiation than LR-4/5 HCCs. In the multivariate analysis, the LR-M category was an independent predictor for early recurrence (hazard ratio, 1.904; 95% confidence interval, 1.024-3.542; P < 0.05). Early recurrence rates were significantly higher in patients with LR-M HCCs than in patients with LR-4/5 HCCs (32.0% vs. 18.4%, respectively, P < 0 05). DATA CONCLUSION: Compared to LR-4/5 HCCs, LR-M HCCs were associated with poorer tumor differentiation and higher early recurrence rates after curative resections of single HCCs. LEVEL OF EVIDENCE: 3 Technical Efficacy Stage: 2.

Diagnostic Performance of Liver Imaging Reporting and Data System Version 2017 Versus Version 2018 for Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis of Comparative Studies.

BACKGROUND: The Liver Imaging Reporting and Data System (LI-RADS) is a comprehensive system for standardizing liver imaging in patients at risk for hepatocellular carcinoma (HCC). PURPOSE: To systematically compare the performance of computed tomography (CT)/MRI LI-RADS category 5 (LR-5) for diagnosing HCC between versions 2017 and 2018. STUDY TYPE: Systematic review and meta-analysis. SUBJECTS: Six articles with 1181 lesions. FIELD STRENGTH/SEQUENCE: 1.5 T and 3.0 T. ASSESSMENT: Data extraction was independently performed by two reviewers who identified and reviewed articles comparing the performance of LR-5 for diagnosing HCC between CT/MRI LI-RADS versions 2017 and 2018. Study and patient characteristics, index test characteristics, reference standards, and study outcomes were extracted from included studies. Risk of bias and concerns regarding applicability were evaluated using the Quality Assessment of Diagnostic Accuracy Studies-2 tool. STATISTICAL TESTS: Bivariate random-effects models were used to calculate the pooled per-observation sensitivity and specificity of LR-5 using both versions. The summary receiver operating characteristic curves were plotted. Meta-regression analysis was performed to explore heterogeneity. A P-value <0.05 was considered to be statistically significant for all analyses other than heterogeneity, where the significance threshold was 0.1. RESULTS: The pooled per-observation sensitivity of LR-5 for diagnosing HCC did not show statistically significant difference between versions 2017 (60%; 95% confidence interval [CI], 49%-70%) and 2018 (67%; 95% CI, 56%-76%; P = 0.381). The pooled per-observation specificities of LR-5 were not significantly different between versions 2017 (92%; 95% CI, 90%-95%) and 2018 (91%; 95% CI, 88%-93%; P = 0.332). Meta-regression analyses revealed that the most common underlying liver disease (hepatitis B or hepatitis C) was a significant factor contributing to the heterogeneity of sensitivities among studies for both versions. DATA CONCLUSION: In this meta-analysis using intraindividual paired comparisons, the pooled sensitivity and pooled specificity of LR-5 were not significantly different between 2017 and 2018 LI-RADS versions. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: Stage 2.

Application of Liver Imaging Reporting and Data System version 2018 ancillary features to upgrade from LR-4 to LR-5 on gadoxetic acid-enhanced MRI.

OBJECTIVES: The current LR-5 criteria of Liver Imaging Reporting and Data System (LI-RADS) determined by only major features provide high specificity, but unsatisfactorily low sensitivity for the noninvasive diagnosis of hepatocellular carcinoma (HCC). This study aimed to identify significant ancillary features (AFs) in LI-RADS version 2018 and develop the upgraded LR-5 criteria to improve diagnostic performance on gadoxetic acid-enhanced magnetic resonance imaging (MRI). METHODS: This retrospective study included 280 patients (366 observations including 281 HCCs) at high-risk for HCC who underwent gadoxetic acid-enhanced MRI between 2015 and 2017. Two readers evaluated major features and AFs for each observation and assigned a LI-RADS category. Independently significant AFs were identified through logistic regression analysis. Upgraded LR-5 criteria were developed by combining independently significant AFs with LR-4 assigned by major features alone. Sensitivities and specificities of the diagnostic criteria were compared using McNemar's test. RESULTS: Two of the AFs favoring malignancy in general (mild-moderate T2 hyperintensity and hepatobiliary phase hypointensity) and two of the AFs favoring HCC in particular (nonenhancing "capsule" and mosaic architecture) were independently significant features for diagnosing HCC. By using the upgraded LR-5 criteria (LR-4 by major features alone + each aforementioned AF), sensitivities were significantly increased (69.4-76.9%) compared with the standard LR-5 (66.2%; all, p ≤ 0.004), whereas specificities (95.3-96.5%) were not significantly different (96.5%; all, p > 0.999). CONCLUSIONS: Independently significant AFs may be used to upgrade from LR-4 to LR-5 to improve sensitivity without impairing specificity on gadoxetic acid-enhanced MRI. KEY POINTS: • Independently significant AFs for HCC on gadoxetic acid-enhanced MRI were mild-moderate T2 hyperintensity, hepatobiliary phase hypointensity, nonenhancing "capsule," and mosaic architecture. • When LR-4 criteria by major features alone in combination with significant AFs were upgraded to LR-5, sensitivities were higher than the standard LR-5, without impairing specificity. • Independently significant ancillary features in Liver Imaging Reporting and Data System version 2018 may be used to upgrade from LR-4 to LR-5 to improve sensitivity without impairing specificity on gadoxetic acid-enhanced MRI.

Contrast-enhanced ultrasound liver imaging reporting and data system for diagnosing hepatocellular carcinoma: A meta-analysis.

BACKGROUND & AIMS: Contrast-enhanced ultrasound (CEUS) Liver Imaging Reporting and Data System (LI-RADS) is a comprehensive system for standardizing CEUS at high risk for hepatocellular carcinoma (HCC). We performed a meta-analysis to determine the diagnostic performance of the CEUS LR-5 for HCC and the pooled proportions of HCCs in each CEUS LI-RADS category. METHODS: We searched multiple databases for studies reporting the diagnostic accuracy of the CEUS LI-RADS. Random-effects model was used to determine summary estimates of the diagnostic performance of CEUS LR-5 and the pooled proportions of HCCs in each CEUS LI-RADS category. Risk of bias and concerns regarding applicability were evaluated with the Quality Assessment of Diagnostic Accuracy Studies-2 tool. RESULTS: Eleven studies were included in the final analysis, which consisted of 5535 observations with 3983 HCCs. The pooled per-observation sensitivity and specificity of the CEUS LR-5 for diagnosing HCC were 69% (95% confidence interval [CI], 64%-73%) and 92% (95% CI, 83%-96%) respectively. The pooled proportions of HCCs were 0% (95% CI, 0-0%) for LR-1, 1% (95% CI, 0%-4%) for CEUS LR-2, 26% (95% CI, 14%-39%) for CEUS LR-3, 77% (95% CI, 68%-86%) for CEUS LR-4, 97% (95% CI, 95%-98%) for CEUS LR-5, 57% (95% CI, 44%-69%) for CEUS LR-M and 100% (95% CI, 93%-100%) for CEUS LR-5V or TIV. CONCLUSIONS: The CEUS LR-5 category showed moderate sensitivity and high specificity for diagnosing HCC. The proportion of HCCs was higher in the higher CEUS LI-RADS categories.

Clinical Results of Transarterial Embolization to Control Postoperative Vascular Complications after Partial Nephrectomy.

PURPOSE: We evaluated the safety and clinical efficacy of transarterial embolization of vascular complications after partial nephrectomy. MATERIALS AND METHODS: This retrospective study included 1,187 patients who underwent partial nephrectomy between January 2006 and December 2017. A total of 36 patients were referred to the interventional radiology department for vascular complications after partial nephrectomy. Data on demographics, clinical manifestations, angiographic findings, the embolization procedure, perioperative details, and technical and clinical success rates were analyzed. Further, renal function was recorded at diagnosis, after embolization and at the last followup. RESULTS: Hemorrhage was diagnosed a median of 5 days (range 0 to 89) postoperatively. The incidence of requiring embolization due to hemorrhage after laparoscopic surgery (5.9% or 17 of 289 cases) was higher than that after open surgery (1.8% or 8 of 440, p = 0.003) and robot-assisted surgery (2.4% or 11 of 458, p = 0.014). The technical and clinical success rates were 100% (36 of 36 patients) and 94.4% (34 of 36) with 2 patients requiring additional embolization with n-butyl-2-cyanoacrylate glue. The mean ± SD estimated glomerular filtration rate at diagnosis, after embolization and at last followup was 81.0 ± 21.6, 83.7 ± 21.0 and 84.9 ± 15.8 ml/minute/1.73 m2, respectively (p = 0.345). No major complication was observed during followup. CONCLUSIONS: Transarterial embolization is safe and effective for managing vascular complications after partial nephrectomy. Moreover, renal function was well preserved with super selective transarterial embolization.

Correction to: Diagnostic efficacy and safety of ultrasound-guided kidney transplant biopsy using cortex-only view: a retrospective single-center study.

The original version of this article, published on 17 December 2018, unfortunately contained a mistake. The following correction has therefore been made in the original: The name of Jaeseung Shin was presented incorrectly. The corrected author list is given above.

Diagnostic efficacy and safety of ultrasound-guided kidney transplant biopsy using cortex-only view: a retrospective single-center study.

PURPOSE: Cortical biopsy is the cornerstone to reveal a cause of unexplained dysfunction of the kidney transplant. Nevertheless, only a few studies have reported the biopsy technique with its performance. We described a novel technique of ultrasound (US)-guided kidney transplant biopsy using cortex-only view and analyzed its diagnostic efficacy and safety. MATERIALS AND METHODS: Between January 2014 and December 2016, a consecutive series of 188 patients who underwent US-guided kidney transplant biopsy using cortex-only view by an experienced radiologist were evaluated (mean age, 46.1 ± 12.5 years; range, 21-79 years). Biopsy time, biopsy distance, biopsy core number, and glomerular number per patient were recorded. Successful biopsy (e.g., adequate, 10 or more glomeruli; marginal, 7-9 glomeruli) and complication rates were investigated, using Banff criteria and Clavien-Dindo classification, respectively. RESULTS: Mean biopsy time, distance, and core number were 20.6 ± 6.7 min (range, 10-44 min), 3.2 ± 0.7 cm (range, 2.1-5.4 cm), and 1.9 ± 0.3 (range, 1.0-3.0), respectively. Mean glomerular number per patient was 20.4 ± 10.0 (range, 0-54). Adequate and marginal biopsy rates were 87.2% (164/188) and 95.2% (179/188), respectively. There was no major complication requiring treatment (no patient with Clavien-Dindo grade 2 or greater complication), while there were self-limiting minor complications in 5 patients (overall complication rate, 2.7%). CONCLUSION: US-guided biopsy using cortex-only view is feasible and safe in sampling cortical tissues of kidney transplant. KEY POINTS: • Ultrasound (US)-guided kidney transplant biopsy using cortex-only view is feasible and safe. • Adequate and marginal biopsy rates were 87.2% and 95.2%, respectively. • No major complication requiring treatment occurred after biopsy.