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

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

LI-RADS category is associated with treatment outcomes of small single HCC: surgical resection vs. radiofrequency ablation.

OBJECTIVES: To assess whether the Liver Imaging Reporting and Data System (LI-RADS) category is associated with the treatment outcomes of small single hepatocellular carcinoma (HCC) after surgical resection (SR) and radiofrequency ablation (RFA). METHODS: This retrospective study included 357 patients who underwent SR (n = 209) or RFA (n = 148) for a single HCC of ≤ 3 cm between 2014 and 2016. LI-RADS categories were assigned. Overall survival (OS), recurrence-free survival (RFS), and local tumor progression (LTP) rates after treatment were compared according to the LI-RADS category (LR-4/5 vs. LR-M) before and after propensity score matching (PSM). Prognostic factors for treatment outcomes were assessed. RESULTS: In total, 357 patients (mean age, 59 years; men, 272) with 357 HCCs (294 LR-4/5 and 63 LR-M) were included. After PSM (n = 78 in each treatment group), there were 10 and 11 LR-M HCCs in the SR and RFA group, respectively. There were no significant differences in OS or RFS. However, SR provided a lower 5-year LTP rate than RFA (1.4% vs. 14.9%, p = 0.001). SR provided a lower 5-year LTP rate than RFA for LR-M HCCs (0% vs. 34.4%, p = 0.062) and LR-4/5 HCCs (1.5% vs. 12.0%, p = 0.008). The LI-RADS category was the sole risk factor associated with poor OS (hazard ratio [HR] 3.79, p = 0.004), RFS (HR 2.12; p = 0.001), and LTP (HR 2.89; p = 0.032). CONCLUSION: LI-RADS classification is associated with the treatment outcome of HCC, supporting favorable outcomes of SR over RFA for LTP, especially for HCCs categorized as LR-M. CLINICAL RELEVANCE STATEMENT: Liver Imaging Reporting and Data System category has a potential prognostic role, supporting favorable outcomes of surgical resection over radiofrequency ablation for local tumor progression, especially for hepatocellular carcinoma categorized as LR-M. KEY POINTS: • SR provided a lower 5-year LTP rate than RFA for HCCs categorized as LR-M (0% vs. 34.4%, p = 0.062) and HCCs categorized as LR-4/5 (1.5% vs. 12.0%, p = 0.008). • There is a steeply increased risk of LTP within 1 year after RFA for LR-M HCCs, compared to SR. • The LI-RADS category was the sole risk factor associated with poor OS (HR 3.79, p = 0.004), RFS (HR 2.12; p = 0.001), and LTP (HR 2.89; p = 0.032) in patients with HCC of ≤ 3 cm treated with SR or RFA.

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.

Interobserver Variability and Diagnostic Performance in Predicting Malignancy of Pancreatic Intraductal Papillary Mucinous Neoplasm with MRI.

Background The 2017 international consensus guidelines for intraductal papillary mucinous neoplasm (IPMN) of the pancreas are widely used. Purpose To evaluate the interobserver agreement and diagnostic performance of MRI assessment in predicting the malignant potential of IPMN according to radiologists' experience. Materials and Methods This multicenter retrospective study included 100 patients with pathologically proven pancreatic IPMN (77 patients with surgery, 23 patients with biopsy) who underwent contrast-enhanced MRI between 2016 and 2021. Eight post-fellowship radiologists (four more-experienced [8-20 years] and four less-experienced [1-4 years] reviewers) evaluated MRI for high-risk stigmata and worrisome features identified by the most recent 2017 guidelines. Interobserver agreement was determined using Fleiss κ statistics according to radiologist experience. The diagnostic performance for malignant IPMN was assessed using receiver operating characteristic curve analysis. Results Among 100 patients (mean age, 66 years ± 10 [SD]; 57 men), 52 (52%) had malignant IPMN. For high-risk stigmata, interobserver agreement was substantial for main pancreatic duct size of at least 10 mm (κ = 0.78; 95% CI: 0.75, 0.82), enhancing mural nodule of at least 5 mm (κ = 0.70: 95% CI: 0.66, 0.74), and at least one high-risk stigmata (κ = 0.73: 95% CI: 0.69, 0.76). The worrisome features showed fair to substantial interobserver agreement (κ range, 0.22-0.80). More-experienced reviewers demonstrated better agreement in the assessment of at least one high-risk stigmata than less-experienced reviewers (κ = 0.77 vs κ = 0.69, P < .001). The overall diagnostic performance of each reviewer was good for the prediction of malignant pancreatic IPMN (area under the receiver operating characteristic curve [AUC] range, 0.77-0.84; median AUC, 0.82), with substantial agreement (κ = 0.76). Conclusion The 2017 international consensus guidelines enabled good diagnostic performance and substantial interobserver agreement for high-risk stigmata but not worrisome features on the evaluation of the malignant pancreatic IPMN using MRI. Agreement tended to be better among more-experienced reviewers than among less-experienced reviewers. © RSNA, 2023 Supplemental material is available for this article.

Differentiation of autoimmune pancreatitis from pancreatic adenocarcinoma using CT characteristics: a systematic review and meta-analysis.

OBJECTIVES: To determine informational CT findings for distinguishing autoimmune pancreatitis (AIP) from pancreatic ductal adenocarcinoma (PDAC) and to review their diagnostic accuracy. METHODS: A systematic and detailed literature review was performed through PubMed, EMBASE, and the Cochrane library. Similar descriptors to embody the identical image finding were labeled as a single CT characteristic. We calculated the pooled diagnostic odds ratios (DORs) of each CT characteristic using a bivariate random-effects model. RESULTS: A total of 145 various descriptors from 15 studies (including 562 AIP and 869 PDAC patients) were categorized into 16 CT characteristics. According to the pooled DOR, 16 CT characteristics were classified into three groups (suggesting AIP, suggesting PDAC, and not informational). Seven characteristics suggesting AIP were diffuse pancreatic enlargement (DOR, 48), delayed homogeneous enhancement (DOR, 46), capsule-like rim (DOR, 34), multiple pancreatic masses (DOR, 16), renal involvement (DOR, 15), retroperitoneal fibrosis (DOR, 13), and bile duct involvement (DOR, 8). Delayed homogeneous enhancement showed a pooled sensitivity of 83% and specificity of 85%. The other six characteristics showed relatively low sensitivity (12-63%) but high specificity (93-99%). Four characteristics suggesting PDAC were discrete pancreatic mass (DOR, 23), pancreatic duct cutoff (DOR, 16), upstream main pancreatic duct dilatation (DOR, 8), and upstream parenchymal atrophy (DOR, 7). CONCLUSION: Eleven CT characteristics were informational to distinguish AIP from PDAC. Diffuse pancreatic enlargement, delayed homogeneous enhancement, and capsule-like rim suggested AIP with the highest DORs, whereas discrete pancreatic mass suggested PDAC. However, pooled sensitivities of informational CT characteristics were moderate. CLINICAL RELEVANCE STATEMENT: This meta-analysis underscores eleven distinctive CT characteristics that aid in differentiating autoimmune pancreatitis from pancreatic adenocarcinoma, potentially preventing misdiagnoses in patients presenting with focal/diffuse pancreatic enlargement. KEY POINTS: • Diffuse pancreatic enlargement (pooled diagnostic odds ratio [DOR], 48), delayed homogeneous enhancement (46), and capsule-like rim (34) were CT characteristics suggesting autoimmune pancreatitis. • The CT characteristics suggesting autoimmune pancreatitis, except delayed homogeneous enhancement, had a general tendency to show relatively low sensitivity (12-63%) but high specificity (93-99%). • Discrete pancreatic mass (pooled diagnostic odds ratio, 23) was the CT characteristic suggesting pancreatic ductal adenocarcinoma with the highest pooled DORs.

Use of individualized 3D-printed models of pancreatic cancer to improve surgeons' anatomic understanding and surgical planning.

OBJECTIVES: Three-dimensional (3D) printing has been increasingly used to create accurate patient-specific 3D-printed models from medical imaging data. We aimed to evaluate the utility of 3D-printed models in the localization and understanding of pancreatic cancer for surgeons before pancreatic surgery. METHODS: Between March and September 2021, we prospectively enrolled 10 patients with suspected pancreatic cancer who were scheduled for surgery. We created an individualized 3D-printed model from preoperative CT images. Six surgeons (three staff and three residents) evaluated the CT images before and after the presentation of the 3D-printed model using a 7-item questionnaire (understanding of anatomy and pancreatic cancer [Q1-4], preoperative planning [Q5], and education for trainees or patients [Q6-7]) on a 5-point scale. Survey scores on Q1-5 before and after the presentation of the 3D-printed model were compared. Q6-7 assessed the 3D-printed model's effects on education compared to CT. Subgroup analysis was performed between staff and residents. RESULTS: After the 3D-printed model presentation, survey scores improved in all five questions (before 3.90 vs. after 4.56, p < 0.001), with a mean improvement of 0.57‒0.93. Staff and resident scores improved after a 3D-printed model presentation (p < 0.05), except for Q4 in the resident group. The mean difference was higher among the staff than among the residents (staff: 0.50‒0.97 vs. residents: 0.27‒0.90). The scores of the 3D-printed model for education were high (trainees: 4.47 vs. patients: 4.60) compared to CT. CONCLUSION: The 3D-printed model of pancreatic cancer improved surgeons' understanding of individual patients' pancreatic cancer and surgical planning. CLINICAL RELEVANCE STATEMENT: The 3D-printed model of pancreatic cancer can be created using a preoperative CT image, which not only assists surgeons in surgical planning but also serves as a valuable educational resource for patients and students. KEY POINTS: • A personalized 3D-printed pancreatic cancer model provides more intuitive information than CT, allowing surgeons to better visualize the tumor's location and relationship to neighboring organs. • In particular, the survey score was higher among staff who performed the surgery than among residents. • Individual patient pancreatic cancer models have the potential to be used for personalized patient education as well as resident education.

Assessment of main pancreatic duct cutoff with dilatation, but without visible pancreatic focal lesion on MDCT: a novel diagnostic approach for malignant stricture using a CT-based nomogram.

OBJECTIVES: To identify useful features to predict hidden pancreatic malignancies in patients with main pancreatic duct (MPD) abrupt cutoff and dilatation, but without visible focal pancreatic lesions on CT. METHODS: This retrospective study included 92 patients (mean age, 63.4 ± 10.6 years, 63 men and 29 women) with MPD abrupt cutoff and dilatation, but without visible focal pancreatic lesion on contrast-enhanced CT between 2009 and 2021. Two radiologists independently evaluated the CT imaging features. Multivariable logistic regression analysis was performed to identify clinical and CT imaging features for hidden pancreatic malignancies. A nomogram was developed based on these results and assessed its performance. RESULTS: Thirty-eight (41.3%) and 54 (58.7%) were classified into the malignant and benign groups, respectively. In the multivariable analysis, CA19-9 elevation (odds ratio [OR] 7.5, p = 0.003), duct cutoff site at the head/neck (OR 7.6, p = 0.006), parenchymal contour abnormality at the duct cutoff site (OR 13.7, p < 0.001), and presence of acute pancreatitis (OR 11.5, p = 0.005) were independent predictors of pancreatic malignancy. A combination of any two significant features showed an accuracy of 77.2%, and a combination of any three features exhibited a specificity of 100%. The CT-based nomogram showed an area under the curve (AUC) of 0.84 (95% confidence interval, 0.77-0.90). CONCLUSIONS: The three CT imaging features and CA19-9 elevation translated into a nomogram permit a reliable estimation of hidden pancreatic malignancies in patients with MPD abrupt cutoff without visible focal pancreatic lesion. It may facilitate determining whether to proceed to further diagnostic tests. KEY POINTS: • Isoattenuating pancreatic ductal adenocarcinoma can manifest only as an isolated main pancreatic duct (MPD) dilatation with abrupt cutoff, making it difficult to distinguish from benign strictures. • Along with the serum CA 19-9 elevation, MPD cutoff site at the pancreas head or neck, parenchymal contour abnormality at the duct cutoff site, and associated acute pancreatitis indicated a higher probability of the malignant MPD strictures. • The CT-based nomogram provided excellent diagnostic performance (AUC of 0.84) for hidden pancreatic malignancies in patients with MPD abrupt cutoff and dilatation.

Differentiating between benign and malignant ampullary strictures: a prediction model using a nomogram based on CT imaging and clinical findings.

OBJECTIVES: To construct a predictive nomogram for differentiating malignant from benign ampullary strictures using contrast-enhanced computed tomography (CT) findings combined with clinical findings. METHODS: In this retrospective study, 152 patients with ampullary stricture (98 benign and 54 malignant) who underwent contrast-enhanced CT were included. Various imaging findings of the ampulla, bile duct, main pancreatic duct, and periampullary area were evaluated and clinical findings including the presence of jaundice, carbohydrate antigen 19-9 level, and history of cholecystectomy were collected. Among them, statistically significant findings were identified using univariable and multivariable logistic regression analyses. A nomogram was constructed to differentiate benign and malignant ampullary strictures and was internally validated. RESULTS: Multivariable analysis revealed that jaundice (odds ratio [OR]: 17.33, p < 0.001), presence of an ampullary mass (OR: 24.40, p < 0.001), non-similar enhancement of the ampulla to the duodenum (OR: 31.96, p = 0.003), and proportional dilatation of the bile duct (OR: 7.98, p = 0.001) were independent significant factors for predicting the malignant ampullary stricture, and were used to construct a nomogram. Among them, non-similar enhancement of the ampulla to the duodenum showed the highest OR and predictor point on the nomogram. The calibration plots showed excellent agreement between the predicted probabilities and the actual rates of malignant ampullary strictures, on internal validation. CONCLUSIONS: Combination of clinical and imaging findings could aid in predicting malignant ampullary strictures using significant findings of jaundice, presence of ampullary mass, non-similar enhancement of the ampulla to the duodenum, and proportional dilatation of the bile duct. KEY POINTS: • The presence of jaundice, ampullary mass, non-similar enhancement of the ampulla, and proportional bile duct dilatation were significant findings for predicting malignant ampullary strictures. • Non-similar enhancement of the ampulla to the duodenum was a significant feature with the highest odds ratio for differentiating benign and malignant ampullary strictures. • The nomogram constructed using contrast-enhanced computed tomography imaging and clinical findings could aid in predicting malignant ampullary strictures.

Deep learning image reconstruction algorithm for abdominal multidetector CT at different tube voltages: assessment of image quality and radiation dose in a phantom study.

OBJECTIVES: To compare the image quality and radiation dose of a deep learning image reconstruction (DLIR) algorithm compared with iterative reconstruction (IR) and filtered back projection (FBP) at different tube voltages and tube currents. MATERIALS AND METHODS: A customized body phantom was scanned at different tube voltages (120, 100, and 80 kVp) with different tube currents (200, 100, and 60 mA). The CT datasets were reconstructed with FBP, hybrid IR (30% and 50%), and DLIR (low, medium, and high levels). The reference image was set as an image taken with FBP at 120 kVp/200 mA. The image noise, contrast-to-noise ratio (CNR), sharpness, artifacts, and overall image quality were assessed in each scan both qualitatively and quantitatively. The radiation dose was also evaluated with the volume CT dose index (CTDIvol) for each dose scan. RESULTS: In qualitative and quantitative analyses, compared with reference images, low-dose CT with DLIR significantly reduced the noise and artifacts and improved the overall image quality, even with decreased sharpness (p < 0.05). Despite the reduction of image sharpness, low-dose CT with DLIR could maintain the image quality comparable to routine-dose CT with FBP, especially when using the medium strength level. CONCLUSION: The new DLIR algorithm reduced noise and artifacts and improved overall image quality, compared to FBP and hybrid IR. Despite reduced image sharpness in CT images of DLIR algorithms, low-dose CT with DLIR seems to have an overall greater potential for dose optimization. KEY POINTS: • Using deep learning image reconstruction (DLIR) algorithms, image quality was maintained even with a radiation dose reduced by approximately 70%. • DLIR algorithms yielded lower image noise, higher contrast-to-noise ratios, and higher overall image quality than FBP and hybrid IR, both subjectively and objectively. • DLIR algorithms can provide a better image quality, much better than FBP and even better than hybrid IR, while facilitating a reduction in radiation dose.

A modified LI-RADS: targetoid tumors with enhancing capsule can be diagnosed as HCC instead of LR-M lesions.

OBJECTIVE: To elucidate whether the presence of enhancing capsule can be applied to establish a modified Liver Imaging Reporting and Data System (LI-RADS) to differentiate hepatocellular carcinoma (HCC) from non-HCC malignancies in extracellular contrast agent (ECA)-enhanced and hepatobiliary agent (HBA)-enhanced MRI. METHODS: We enrolled 198 participants (161 men; mean age, 56.3 years) with chronic liver disease who underwent ECA-MRI and HBA-MRI before surgery for de novo hepatic nodule(s). Two reviewers assigned LI-RADS categories (v2018). We defined a "modified LR-5 category, which emphasizes enhancing capsule (mLR-5C)" over targetoid features and classifies tumors with both targetoid appearance and enhancing capsule as HCC instead of LR-M. We compared the diagnostic performance of conventional LI-RADS and modified LI-RADS criteria for both MRIs. RESULTS: A total of 258 hepatic nodules (194 HCCs, 43 benign lesions, and 21 non-HCC malignancies; median size, 19 mm) were analyzed. By conventional LI-RADS, 47 (18.2%) nodules (31 HCCs and 16 non-HCC malignancies) were categorized as LR-M. The mLR-5C criterion showed superior sensitivity (ECA-MRI, 76.6% vs. 67.0%; HBA-MRI, 60.4% vs. 56.3%; both p < 0.05) while maintaining high specificity (ECA-MRI, 93.8% vs. 98.4%; HBA-MRI, 95.3% vs. 98.4%; both p > 0.05) compared with the LR-5 criterion. Using the mLR-5C criterion, ECA-MRI exhibited higher sensitivity than HBA-MRI (76.6% vs. 60.4%, p < 0.001) and similar specificity (93.8% vs. 95.3%, p > 0.99). CONCLUSION: Our modified LI-RADS achieved superior sensitivity for diagnosing HCC, without compromising specificity compared with LR-5. ECA-MRI showed higher sensitivity in diagnosing HCC than HBA-MRI by applying the mLR-5C for LR-M lesions. KEY POINTS: • By conventional LI-RADS, 31 (16.0%) of 194 HCCs were categorized as LR-M. • Among 31 HCCs categorized as LR-M, 19 HCCs or 8 HCCs were recategorized as HCC on ECA-MRI or HBA-MRI, respectively, after applying the modified LR-5 category, which allocates targetoid lesions with enhancing capsule as mLR-5C instead of LR-M. • The mLR-5C showed superior sensitivity compared with the LR-5 in both MRIs (ECA-MRI, 76.6% vs. 67.0%; HBA-MRI, 60.4% vs. 56.3%, both p < 0.05), while maintaining high specificity (ECA-MRI, 93.8% vs. 98.4%; HBA-MRI, 95.3% vs. 98.4%; both p > 0.05).

Utility of Diffusion-Weighted MRI for Detection of Locally Recurrent Pancreatic Cancer After Surgical Resection.

BACKGROUND. Overlapping imaging findings between local tumor recurrence and postsurgical fibrosis represent a major clinical challenge after pancreatic ductal adeno-carcinoma (PDAC) resection. OBJECTIVE. The purpose of this study was to compare the diagnostic performance of MRI with and without DWI for differentiating locally recurrent tumor and postsurgical fibrosis after PDAC resection. METHODS. This retrospective study included 66 patients (35 men, 31 women; mean age, 60.5 years) who underwent PDAC resection between January 2009 and March 2016, postoperative surveillance CT showing a soft-tissue lesion at the operative site or at the site of peripancreatic vessels, and subsequent MRI with DWI for further evaluation. CT at least 6 months after MRI served as the reference standard, with increase in size of the soft tissue by 5 mm or more differentiating locally recurrent tumor (n = 26) and postsurgical fibrosis (n = 40). Two observers in consensus evaluated MRI characteristics of the soft-tissue lesions. Two additional observers independently reviewed MRI examinations in two separate sessions (conventional MRI alone vs MRI with DWI), recording likelihood of recurrent tumor using a 1-5 scale. ROC analysis was performed, considering scores of 4 or 5 as positive. RESULTS. Subjective diffusion restriction was more common in locally recurrent tumor than postsurgical fibrosis (88.5% vs 25.0%, p = .01). Median ADC was lower for locally recurrent tumor than postsurgical fibrosis (1.3 vs 1.7 × 10-3 mm2/s, p < .001). For both observers, MRI with DWI in comparison with conventional MRI alone showed higher AUC for diagnosis of locally recurrent tumor (observer 1: 0.805 vs 0.707, p = .048; observer 2: 0.898 vs 0.637, p < .001) and higher sensitivity (observer 1: 88.5% vs 61.5%, p = .008; observer 2: 84.6% vs 42.3%, p = .001) but no difference in specificity (observer 1: 72.5% vs 80.0%, p = .08; observer 2, 95.0% vs 85.0%, p = .10). Interobserver agreement was moderate for conventional MRI (κ = 0.41) and good for conventional MRI with DWI (κ = 0.62). CONCLUSION. The addition of DWI to conventional MRI improves the differentiation of locally recurrent tumor and postsurgical fibrosis after PDAC resection, primarily because of improved sensitivity for recurrence. CLINICAL IMPACT. The findings indicate a potential role for MRI with DWI in surveillance protocols after PDAC resection.

LI-RADS Version 2018 Targetoid Appearances on Gadoxetic Acid-Enhanced MRI: Interobserver Agreement and Diagnostic Performance for the Differentiation of HCC and Non-HCC Malignancy.

BACKGROUND. In LI-RADS version 2018, observations showing at least one of five targetoid appearances in different sequences or postcontrast phases are categorized LR-M, indicating likely non-hepatocellular carcinoma (HCC) malignancy. OBJECTIVE. The purpose of this study was to evaluate interobserver agreement for LI-RADS targetoid appearances among a large number of radiologists of varying experience and the diagnostic performance of targetoid appearances for differentiating HCC from non-HCC malignancy. METHODS. This retrospective study included 100 patients (76 men, 24 women; mean age, 58 ± 9 [SD] years) at high risk of HCC who underwent gadoxetic acid-enhanced MRI within 30 days before hepatic tumor resection (25 randomly included patients with non-HCC malignancy [13, intrahepatic cholangiocarcinoma; 12, combined HCC-cholangiocarcinoma]; 75 matched patients with HCC). Eight radiologists (four more experienced [8-15 years]; four less experienced [1-5 years]) from seven institutions independently assessed observations for the five targetoid appearances and LI-RADS categorization. Interobserver agreement and diagnostic performance for non-HCC malignancy were evaluated. RESULTS. Interobserver agreement was poor for peripheral washout (κ = 0.20); moderate for targetoid transitional phase or hepatobiliary phase appearance (κ = 0.33), delayed central enhancement (κ = 0.37), and targetoid restriction (κ = 0.43); and substantial for rim arterial phase hyperenhancement (κ = 0.61). Agreement was fair for at least one targetoid appearance (κ = 0.36) and moderate for at least two, three, or four targetoid appearances (κ = 0.43-0.51). Agreement for individual targetoid appearances was not significantly different between more experienced and less experienced readers other than for targetoid restriction (κ = 0.63 vs 0.43; p = .001). Agreement for at least one targetoid appearance was fair among more experienced (κ = 0.29) and less experienced (κ = 0.37) reviewers. Agreement for at least two, three, or four targetoid appearances was moderate to substantial among more experienced reviewers (κ = 0.45-0.63) and moderate among less experienced reviewers (κ = 0.42-0.56). Existing LR-M criteria of at least one targetoid appearance had median accuracy for non-HCC malignancy of 62%, sensitivity of 84%, and specificity of 54%. For all reviewers, accuracy was highest when at least three (median accuracy, 79%; sensitivity, 68%; specificity, 82%) or four (median accuracy, 80%; sensitivity, 54%; specificity, 88%) targetoid appearances were required. CONCLUSION. Targetoid appearances and LR-M categorization exhibited considerable interobserver variation among both more and less experienced reviewers. CLINICAL IMPACT. Requiring multiple targetoid appearances for LR-M categorization improved interobserver agreement and diagnostic accuracy for non-HCC malignancy.

Laparoscopic radiofrequency ablation of subcapsular hepatocellular carcinomas: risk factors related to a technical failure.

BACKGROUND: This study aimed to evaluate the risk factors related to a technical failure after laparoscopic radiofrequency ablation (RFA) for subcapsular hepatocellular carcinomas (HCCs). MATERIALS AND METHODS: A total of 110 patients with 114 HCCs who underwent laparoscopic RFA for HCCs (new HCC [n = 85] and local tumor progression [LTP] [n = 29]) between January 2013 and December 2018 were included. We evaluated the incidence of technical failure on immediate post-RFA CT images. Risk factors for a technical failure after laparoscopic RFA were assessed using univariable logistic regression analyses. The cumulative LTP rate was estimated using the Kaplan-Meier method. RESULTS: Technical failure was noted in 3.5% (4/114) of the tumors. All four tumors that showed a technical failure were cases of LTP from previous treatment and were invisible on laparoscopy. On univariate analysis, LTP lesion, invisibility of the index tumor on laparoscopy, and peri-hepatic vein location of the tumor were identified as risk factors for a technical failure. The cumulative LTP rates at 1, 3, and 5 years were estimated to be 2.8%, 4.8%, and 4.8%, respectively. CONCLUSIONS: LTP lesion, invisibility of the index tumor on laparoscopy, and peri-hepatic vein location of the tumor were identified as the risk factors for a technical failure after laparoscopic RFA.

Comparison of Super-Resolution US and Contrast Material-enhanced US in Detection of the Spoke Wheel Sign in Patients with Focal Nodular Hyperplasia.

Background Diagnosis of focal nodular hyperplasia (FNH) with US generally requires the use of contrast material. The effect of the super-resolution US technique on the diagnosis of FNH is unknown. Purpose To investigate the equivalence between super-resolution US and contrast material-enhanced US in the detection of spoke wheel sign in patients with FNH by comparing patterns of tumor vascularity. Materials and Methods This is a secondary analysis of a prospective trial (NCT02737865) that enrolled participants diagnosed with FNH between May 2016 and March 2019. These patients underwent super-resolution US and subsequent contrast-enhanced US with perfluorobutane microbubbles on the same day. The primary outcome was the confidence score of detecting spoke wheel sign in patients with FNH at US. Two radiologists used a four-point scale to score their confidence in the presence of the spoke wheel sign based on super-resolution US and contrast-enhanced US findings. Two one-sided tests were used to test the equivalence between super-resolution US and contrast-enhanced US in terms of the score for the confidence level of the spoke wheel sign. Interobserver agreement for both techniques between the two radiologists, using the recorded images, was analyzed by using an intraclass correlation coefficient. Results In 62 patients (mean age, 37 years; range, 20-69 years; 41 women) with FNH, the majority of patients showed a spoke wheel sign at super-resolution US and contrast-enhanced US (63% [39 of 62] and 71% [44 of 62], respectively; P = .36). There was no significant difference between the super-resolution US and contrast-enhanced US techniques regarding the confidence score for the spoke wheel sign (mean score, 1.8 vs 2.0; P = .03 for equivalence test). The intraclass correlation coefficients of super-resolution US and contrast-enhanced US regarding the presence of the spoke wheel sign were 0.82 (95% confidence interval: 0.73, 0.96) and 0.58 (95% confidence interval: 0.41, 0.73), respectively. Conclusion In comparison with contrast-enhanced US, super-resolution US provided a reliable rate of detection of the spoke wheel sign in patients with focal nodular hyperplasia. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Fetzer in this issue.

Abbreviated Magnetic Resonance Imaging With Breath-Hold Three-Dimensional Magnetic Resonance Cholangiopancreatography: Assessment of Malignant Risk of Pancreatic Intraductal Papillary Mucinous Neoplasm.

BACKGROUND: For surveillance of pancreatic intraductal papillary mucinous neoplasms (IPMNs), magnetic resonance imaging (MRI)/magnetic resonance cholangiopancreatography (MRCP) is preferred over computed tomography, but the long acquisition time limits its use. PURPOSE: To investigate the diagnostic performance of abbreviated MRI with breath-hold (BH) three-dimensional MRCP (abbreviated MRI-BH) for malignant risk assessment of pancreatic IPMN. STUDY TYPE: Retrospective. POPULATION: Two hundred and thirty-five patients with IPMNs (M:F = 115:120; mean age ± SD, 66 ± 9 years; typical imaging features with ≥2-year stability [N = 172] and histopathologically confirmed [N = 63]). FIELD STRENGTH/SEQUENCE: 3.0 T/ abbreviated MRI-BH (single-shot fast spin-echo, T1W fat-suppressed gradient-echo sequence, and BH-3D-MRCP). ASSESSMENT: Abbreviated MRI-BH was reviewed by three reviewers, and its diagnostic performance was assessed using the predetermined scoring system. The diagnostic performance for the mural nodule detection was assessed. Additionally, diagnostic performance of abbreviated MRI was compared with that of full-sequence MRI. STATISTICAL TESTS: Area under the receiver operating characteristic curve (AUC) with z-test, and linear-weighted kappa values. RESULTS: Thirty-five patients had malignant IPMN. At a cut-off score ≥3, AUCs of abbreviated MRI-BH for detecting malignant IPMN were 0.959 for reviewer 1, 0.962 for reviewer 2, and 0.956 for reviewer 3. The sensitivity of reviewers 1, 2, and 3 was 97.1% for all, and the specificity was 85.5%, 86.0%, and 85.0%, respectively. Regarding mural nodule detection (N = 22), abbreviated MRI-BH demonstrated a sensitivity of 95.5% and a specificity of 88.3% for reviewer 1, a sensitivity of 86.4% and a specificity of 92.0% for reviewer 2, and a sensitivity of 86.4% and a specificity of 89.2% for reviewer 3. There were no significant differences between AUC of abbreviated MRI-BH and that of full-sequence MRI in the three reviewers (P > 0.05). DATA CONCLUSION: Abbreviated MRI-BH showed good diagnostic performance for detecting malignant IPMNs by using a predetermined scoring system. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY STAGE: 2.

EASL versus LI-RADS: Intra-individual comparison of MRI with extracellular contrast and gadoxetic acid for diagnosis of small HCC.

BACKGROUND & AIMS: Liver Imaging Reporting and Data System (LI-RADS) and European Association for the Study (EASL) criteria for hepatocellular carcinoma (HCC) diagnosis have been updated in 2018. We aimed to compare the HCC diagnostic performance of LI-RADS and EASL criteria with extracellular contrast agents-MRI (ECA-MRI) and hepatobiliary agents-MRI (HBA-MRI). METHODS: We prospectively evaluated 179 participants with cirrhosis (n = 105) or non-cirrhotic chronic hepatitis B (CHB) (n = 74) who underwent both ECA-MRI and HBA-MRI before surgery for de novo nodule(s) measuring 10-30 mm. We compared the HCC diagnostic performance of EASL and LR-5 in both MRIs. RESULTS: In an analysis of 215 observations (175 HCCs, 17 non-HCC malignancies and 23 benign lesions) identified from cirrhotic or non-cirrhotic CHB participants, LR-5 with ECA-MRI provided the highest sensitivity (80.7%), followed by EASL with ECA-MRI (76.2%), LR-5 with HBA-MRI (67.3%) and EASL with HBA-MRI (63.0%, all P < .05). The specificities were comparable (89.4%-91.5%). When the analysis is limited to participants with pathological cirrhosis (123 observations), the sensitivity of LR-5 with ECA-MRI was similar to that of EASL with ECA-MRI (82.7% vs 80.2%, P = .156), but higher than LR-5 with HBA-MRI (65.1%) or EASL with HBA-MRI (62.8%, both P < .001), with comparable specificities (87.5%-91.7%). CONCLUSIONS: The LR-5 with ECA-MRI yielded the highest sensitivity with a similar specificity for HCC diagnosis in cirrhosis and non-cirrhotic CHB participants, while the sensitivities of LR-5 and EASL with ECA-MRI are similar for cirrhosis participants. This indicates non-invasive diagnosis criteria can differ by contrast agents and presence of cirrhosis.

Intraductal papillary mucinous neoplasm of the pancreas: diagnostic performance of the 2017 international consensus guidelines using CT and MRI.

OBJECTIVES: To assess the diagnostic performance of the 2017 international consensus guidelines for intraductal papillary mucinous neoplasm (IPMN) of the pancreas and to compare the diagnostic performance and intermodality agreement between contrast-enhanced CT and MRI. METHODS: We retrospectively evaluated patients with surgical resection of IPMN of the pancreas who underwent preoperative CT and MRI between 2009 and 2019. Two radiologists evaluated the clinical and imaging features of IPMN of pancreas according to the 2017 international consensus guideline. Univariable and multivariable analyses were performed to identify significant predictors of malignancy in IPMN. The diagnostic abilities of CT and MRI were compared, and their intermodality agreement was determined. RESULTS: Of 175 patients (mean age, 64 years; 116 males), 88 (50.3%) had malignant IPMN. On multivariable analysis, all three high-risk stigmata (main pancreatic duct [MPD] ≥ 10 mm, mural nodule ≥ 5 mm, and obstructive jaundice) and two worrisome features (MPD 5-9 mm and elevated carbohydrate antigen 19-9) were associated with malignant IPMN on CT and MRI (p < 0.05). A mural nodule < 5 mm on MRI was also associated with malignant IPMN (OR 5.3, p = 0.009). The diagnostic accuracy of high-risk stigmata showed no difference between CT and MRI (73.7% vs. 75.4%, p = 0.505), with good to excellent intermodality agreement. CONCLUSIONS: Current high-risk stigmata had the strongest association with malignant IPMN on CT and MRI. Although MRI is superior to CT for identifying mural nodules, diagnostic performance for differentiating malignant from benign IPMNs was similar between CT and MRI. KEY POINTS: • The current high-risk stigmata in the 2017 International Consensus Guidelines had the strongest association with malignant IPMN on CT and MRI. • MRI is better than CT for identifying enhancing mural nodule. • Diagnostic performance for differentiating malignant from benign IPMNs was similar between CT and MRI.

Assessment of factors affecting washout appearance of hepatocellular carcinoma on CT.

OBJECTIVES: To identify independent imaging and histopathologic factors that affect washout appearance of hepatocellular carcinoma (HCC) in CT images. METHODS: This retrospective study included 264 patients who had undergone surgical resection for treatment-naïve single HCC between January 2014 and December 2015 and had available preoperative multiphasic CT images. Two reviewers evaluated the CT imaging features of HCC using LI-RADS v2018. The "washout" was visually assessed in portal venous or equilibrium phases. Depending on the presence of washout appearance of HCC, all patients were divided into "washout" (n = 228) and "no washout" (n = 36) groups. Multivariable logistic regression analysis was used to identify factors associated with the absence of washout appearance of HCC. RESULTS: A total of 264 HCCs (median size, 2.6 cm) were analyzed. Histologically proven hepatic steatosis (macrovesicular steatosis ≥ 5%) (odds ratio [OR], 2.65; 95% confidence interval [CI], 1.05-6.74; p = 0.040), tumor capsule on histopathology (OR, 0.17; 95% CI, 0.06-0.50; p = 0.001), and mosaic appearance on CT image (OR, 0.34; 95% CI, 0.14-0.85; p = 0.021) were independent factors associated with the absence of washout appearance of HCC. In 189 patients with available unenhanced CT images, CT-diagnosed hepatic steatosis was also an independent factor for the absence of washout appearance of HCC (OR, 9.26; 95% CI, 3.06-28.02; p < 0.001). CONCLUSIONS: Washout appearance of HCC in CT images could be obscured in both histologically proven hepatic steatosis and CT-diagnosed hepatic steatosis, and could be enhanced with tumor capsule on histopathology and mosaic appearance on CT image. KEY POINTS: • Hepatic steatosis is an independent factor related to the absence of washout appearance of hepatocellular carcinoma in CT images, in both histologically proven hepatic steatosis and CT-diagnosed hepatic steatosis. • Both histologically proven hepatic steatosis and CT-diagnosed hepatic steatosis have higher odds of absence of washout appearance of hepatocellular carcinoma compared to non-steatotic liver. • Tumor capsule on histopathology and mosaic appearance on CT image are independent factors that enhance the probability that washout appearance of hepatocellular carcinoma is visible.

Vanishing washout of hepatocellular carcinoma according to the presence of hepatic steatosis: diagnostic performance of CT and MRI.

OBJECTIVES: To compare the presence of washout and the diagnostic performance of computed tomography (CT) and magnetic resonance imaging (MRI) for hepatocellular carcinoma (HCC) according to the presence of hepatic steatosis. METHODS: This retrospective study included 566 patients with chronic liver disease who had undergone hepatic resection for hepatic tumors (482 HCCs and 84 non-HCCs) between January 2016 and June 2018 and had available multiphasic CT and MR images. Patients were allocated in the fatty liver (n = 141) or non-fatty liver (n = 425) group according to the presence of hepatic steatosis, defined as lipid droplets in at least 5% of hepatocytes on pathological examination. The presence of HCC washout and the diagnostic performance of CT and MRI for HCC were compared between the groups. RESULTS: HCC washout was less frequently seen in the fatty liver group than in the non-fatty liver group on CT (61.5% vs. 88.9%, p < 0.001), whereas it was similarly present on MRI in both groups (77.0% vs. 74.4%, p = 0.565). For diagnosis of HCC, the sensitivity (53.3% vs. 80.0%, p < 0.001) and accuracy (53.9% vs. 80.9%, p < 0.001) of CT were lower in the fatty liver group than in the non-fatty liver group. However, for MRI, these values were not significantly different between the groups (p > 0.05). CONCLUSIONS: Hepatic steatosis significantly decreased the performance of CT for the diagnosis of HCC, whereas it did not significantly alter the performance of MRI. KEY POINTS: • Unlike MRI, there is vanishing HCC washout on CT caused by the background hepatic steatosis. • The diagnostic performance of CT for the diagnosis of HCC was significantly altered by hepatic steatosis. • The optimal cutoff HU value of the liver parenchyma for the vanishing washout of HCC was < 50 HU on unenhanced CT images.

Rim-arterial enhancing primary hepatic tumors with other targetoid appearance show early recurrence after radiofrequency ablation.

OBJECTIVES: To evaluate early (≤ 2 years) local tumor progression (LTP), intrahepatic distant metastasis (IDR), and extrahepatic metastasis (EM) of primary hepatic malignant tumors with arterial rim enhancement (RE) after RFA in comparison with non-RE tumors. METHODS: Three hundred forty-nine patients who underwent RFA for primary hepatic malignant tumors between January 2009 and December 2016 were included. The patients' tumors were classified into non-RE, RE only (RO), and RE plus other targetoid appearances (REoT). Cumulative LTP, IDR, and EM rates at 1 and 2 years after RFA were calculated using the Kaplan-Meier method and compared using the log-rank test. Prognostic factors for the outcomes were assessed using a Cox proportional hazards model. RESULTS: There were 303 non-RE, 19 RO, and 27 REoT tumors. The REoT tumors had a significantly higher rate of IDR and EM than non-RE (p = 0.04 for IDR; and p < 0.01 for EM, respectively) at 1 year after RFA. At 2 years, LTP and EM rates were significantly higher for REoT than for non-RE (p = 0.001 for LTP; and p = 0.444 for EM, respectively). The RO tumors did not have different outcomes than non-RE at 1 and 2 years after RFA. Multivariable analysis verified that REoT was a significant factor for IDR (p = 0.04) and EM (p = 0.01) at 1 year and LTP (p = 0.02) at 2 years. CONCLUSIONS: Tumors with REoT had poor LTP, IDR, and EM within 2 years after RFA than non-RE tumors. However, tumors with RO showed similar results as non-RE tumors. KEY POINTS: • Tumors with Rim enhancement plus other targetoid appearances (REoT) had a significantly higher rate of recurrence than non-rim enhancing (RE) tumors at 1 and 2 years after RFA. • Tumors with rim enhancement only did not have different outcomes than non-RE at 1 and 2 years after RFA.