Unique endoscopic features of primary biliary diffuse large B-cell lymphoma: A case report with literature review (with video).

A 67-year-old man visited our hospital complaining of dark-colored urine and upper abdominal pain. Magnetic resonance cholangiopancreatography showed stricture of the distal bile duct, and contrast-enhanced computed tomography showed irregular thickening of the distal bile duct wall. However, no enlarged lymph nodes, pancreatic tumors, or other neoplastic lesions were apparent around the bile duct. Endoscopic ultrasonography and intraductal ultrasonography showed irregular thickening of the inner hypoechoic layer without the disappearance of the innermost thin hyperechoic layer. On the basis of these findings, we considered that the bile duct lesion was of non-epithelial origin. Thus, we repeatedly performed bile duct biopsies from the same site under fluoroscopy to obtain a sample of the submucosal tissue. The pathological diagnosis was diffuse large B-cell lymphoma, and the patient received systemic chemotherapy (rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone). After six courses of rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone, positron emission tomography-computed tomography showed the disappearance of 18-fluorodeoxyglucose uptake in the bile duct and endoscopic retrograde cholangiography showed improvement of the bile duct stricture. Endoscopic findings and repeated biopsies were useful in making the diagnosis of primary biliary diffuse large B-cell lymphoma.

Performance evaluation of endoscopic ultrasonography needles: Experimental study.

Background and study aims The relative procedural performance of needles for endoscopic ultrasound-guided fine-needle aspiration/biopsy (EUS-FNA/B) is unclear. The present study therefore compared six types of 22-gauge FNA/B needles using a bench simulator. Methods Resistance forces during needle puncture and removal, needle tip damage before and after procedure, leakage after puncture of mucinous cyst models, the shape of the puncture surface at the puncture site, amounts of samples extracted, ranges of needle deflection angles, and needle deformation after multiple procedures were compared using six types of needles. Results Maximum resistance forces during puncture and removal were highest for ProCore needles and lowest for Expect needles. None of the needles had damage after puncturing. SharkCore needles showed the highest amount of leakage, whereas FNA needles showed no leakage. The puncture tracts of FNA needles remained in the form of a flap at the puncture site, whereas FNB needles broke off the target material creating a hole. The target material removed was supplemented within the puncture needle. TopGain needles produced significantly larger samples than ProCore, EZShot3 Plus, and Expect needles. FNB needles produced larger and more core samples than FNA needles. EZShot3 Plus needles had the highest range of needle deflection angle using an elevator device and the lowest needle deformation after 20 punctures at full endoscopic angle and a full elevator. Conclusions The performance of the six needles differed in various ways. Understanding the characteristics of each needle may allow for selection of the appropriate needle for each situation.

Misdiagnosis of hemangioma of left triangular ligament of the liver as gastric submucosal stromal tumor: Two case reports.

BACKGROUND: Extragastric lesions are typically not misdiagnosed as gastric submucosal tumor (SMT). However, we encountered two rare cases where extrinsic lesions were misdiagnosed as gastric SMTs. CASE SUMMARY: We describe two cases of gastric SMT-like protrusions initially misdiagnosed as gastric SMTs by the abdominal contrast-enhanced computed tomography (CT) and endoscopic ultrasound (EUS). Based on the CT and EUS findings, the patients underwent gastroscopy; however, no tumor was identified after incising the gastric wall. Subsequent surgical exploration revealed no gastric lesions in both patients, but a mass was found in the left triangular ligament of the liver. The patients underwent laparoscopic tumor resection, and the postoperative diagnosis was hepatic hemangiomas. CONCLUSION: During EUS procedures, scanning across different layers and at varying degrees of gastric cavity distension, coupled with meticulous image analysis, has the potential to mitigate the likelihood of such misdiagnoses.

Endoscopic ultrasonography for microvascular imaging without contrast enhancement in the differential diagnosis of pancreatic lesions.

OBJECTIVES: Detective flow imaging endoscopic ultrasonography (DFI-EUS) is a recent imaging modality developed for detecting fine vessels without the need for ultrasound contrast agents. The aim of the present study was to evaluate the utility of DFI-EUS for solid pancreatic lesions and to compare the diagnostic ability for pancreatic cancer (PC) between DFI-EUS, directional power Doppler (eFLOW) EUS, and contrast-enhanced harmonic (CH)-EUS. METHODS: Patients with a pancreatic lesion who underwent DFI-EUS, eFLOW-EUS, and CH-EUS between March 2019 and November 2023 were retrospectively enrolled. Final diagnoses were confirmed by pathologic examination of EUS-guided tissue acquisition and/or resected specimens. Lesions were categorized into the three patterns of poor, mild, and rich vascularity on DFI-EUS and eFLOW-EUS, and hypo-, iso-, and hypervascular on CH-EUS. PC was defined as a poor pattern on DFI-EUS and eFLOW-EUS, and a hypovascular pattern on CH-EUS. RESULTS: The final diagnoses of 90 examined tumors were PC (n = 57), inflammatory mass (n = 6), autoimmune pancreatitis (n = 13), neuroendocrine tumor (n = 9), and others (n = 5). The sensitivity, specificity, and accuracy for diagnosis of PC were 93%, 82%, and 88%, respectively, on DFI-EUS, 97%, 42%, and 77% on eFLOW-EUS, and 95%, 89%, and 92% on CH-EUS. The accuracy of DFI-EUS was significantly superior to eFLOW-EUS (P = 0.005), but no significant difference was found between DFI-EUS and CH-EUS. CONCLUSION: DFI-EUS is more sensitive for depicting vasculature than eFLOW-EUS, and has higher diagnostic sensitivity for PC. Evaluation of vascularity on DFI-EUS is useful for the differential diagnosis of pancreatic lesions without the need for intravenous contrast agent.

A novel endoscopic ultrasomics-based machine learning model and nomogram to predict the pathological grading of pancreatic neuroendocrine tumors.

Objectives: This research aimed to retrospectively construct and authenticate ultrasomics models using endoscopic ultrasonography (EUS) images for forecasting the pathological grading of pancreatic neuroendocrine tumors (PNETs). Methods: After confirmation through pathological examination, a retrospective analysis of 79 patients was conducted, including 49 with grade 1 PNETs and 30 with grade 2/3 PNETs. These patients were randomized to the training or test cohort in a 6:4 proportion. The least absolute shrinkage and selection operator (LASSO) algorithm was used to reduce the dimensionality of ultrasomics features derived from standard EUS images. These nonzero coefficient features were retained and applied to construct prediction models via eight machine-learning algorithms. The optimum ulstrasomics model was determined, followed by creating and evaluating a nomogram. Results: Ultrasomics features of 107 were extracted, and only those with coefficients greater than zero were retained. The XGboost ultrasomics model performed exceptionally well, achieving AUCs of 0.987 and 0.781 in the training and test cohorts, respectively. Furthermore, an effective nomogram was developed and visually represented. Finally, the calibration curves, decision curve analysis (DCA) plots, and clinical impact curve (CIC) displayed in the ulstrasomics model and nomogram demonstrated high accuracy. They provided significant net benefits for clinical decision-making. Conclusions: A novel ulstrasomics model and nomogram were created and certified to predict the pathological grading of PNETs using EUS images. This study has the potential to provide valuable insights that improve the clinical applicability and efficacy of EUS in predicting the grading of PNETs.

Endoscopic ultrasound with tissue acquisition of lymph nodes in patients with potentially resectable intrahepatic cholangiocarcinoma.

Background and study aims Lymph node (LN) involvement is a poor prognostic factor for patients with intrahepatic cholangiocarcinoma (iCCA). The aim of this study was to evaluate the yield and impact on clinical decision making of endoscopic ultrasound with tissue acquisition (EUS-TA) of LNs in patients with potentially resectable iCCA. Patients and methods In this multicenter cohort study, patients with potentially resectable iCCA and preoperative EUS between 2010 and 2020 were retrospectively included. The impact of EUS-TA was defined as the percentage of patients who did not undergo surgical exploration due to pathologically confirmed positive LNs found with EUS-TA. Results A total of 56 patients underwent EUS, with 91% of patients to target suspicious LNs on imaging. EUS-TA of LNs confirmed malignancy in 21 LNs among 19 patients (34%). In 17 patients (30%), surgical exploration was withheld due to nodal involvement. Finally, 24 patients (43%) underwent surgical exploration among whom positive regional LNs were identified in six patients (25%). Conclusions In patients with potentially resectable iCCA and suspicious LNs on cross-sectional imaging, EUS-TA confirmed LN involvement in 30% of patients. Surgical exploration was withheld mostly because of extraregional LN involvement and regional LN involvement in patients with high surgical risk.

Mortality from upper gastrointestinal tumors in colorectal cancer screening patients.

Background and study aims Currently, gastric cancer screening is only cost-effective in countries with high incidence. Integrated screening, in which gastroscopy is performed in conjunction with colonoscopy, could help reduce the gastric cancer screening procedure burden in countries with low or intermediate incidence. However, there is a lack of population-based studies to identify high-risk groups. Methods In this retrospective analysis of a colorectal cancer (CRC) screening program database, we used Cox proportional hazards model to identify an association of high- and low-risk finding (polyps ≥ 10 mm or with high-grade dysplasia) with time to death from upper gastrointestinal cancer (esophageal and gastric). We estimated the 10-year mortality of upper gastrointestinal tumors in different 10-year age groups, stratified by sex and polyp finding at colonoscopy. Results We included 349,856 CRC screening colonoscopies in our study. The median follow-up time was 5.22 years (95% confidence interval [CI] 5.21-5.24 years). Of the participants, 4.5% had polyps ≥ 10 mm or with high-grade dysplasia (HGD). At the end of the study period, 384 deaths from upper gastrointestinal cancer had occurred. Aside from age and sex, we found the presence of high-risk polyps to be significantly associated with upper gastrointestinal cancer death (hazard ratio 1.54, 95% CI 1.06-2.25, P = 0.025). Conclusions CRC screening participants with polyps < 10 mm and no HGD have a lower risk for mortality from upper gastrointestinal cancers compared with participants with polyps > 10 mm and HGD. Future studies will demonstrate whether integrated screening with additional gastroscopy is effective in CRC screening participants with large or highly dysplastic polyps.

A Case of Gastric Leiomyosarcoma Completely Resected by Endoscopic Submucosal Dissection after a Precise Preoperative Diagnosis.

A 68-year-old woman was diagnosed with leiomyosarcoma (LMS) based on preoperative biopsy of the gastric body. As tumor invasion confined to the submucosa with no breaking of the submucosal layer was confirmed on endoscopic ultrasonography (EUS), the patient underwent endoscopic submucosal dissection (ESD) for gastric LMS, resulting in complete tumor resection. No apparent recurrence was observed in the 2.5 years after treatment. This is an extremely rare case of gastric LMS that underwent ESD after a precise preoperative diagnosis, with no signs of recurrence after treatment. ESD may be an acceptable option for gastric LMS when EUS findings allow this treatment method.

Endoscopic resection in subepithelial lesions of the upper gastrointestinal tract: Experience at a tertiary referral hospital in The Netherlands.

Background and study aims Histological confirmation of subepithelial lesions (SELs) in the upper gastrointestinal tract remains challenging. Endoscopic resection of SELs is increasingly used for its excellent diagnostic yield and opportunity to do away with continued surveillance. In this study, we aimed to evaluate the indications, success rates and complications of different endoscopic resection techniques for SELs at a large, tertiary referral hospital in Rotterdam, The Netherlands. Patients and methods Data between October 2013 and December 2021 were retrospectively collected and analyzed. Main outcomes were R0-resection rate, en bloc resection rate, recurrence rate, and procedure-related adverse events (AEs) (Clavien-Dindo). Secondary outcomes were procedure time, need for surgical intervention, and clinical impact on patient management. Results A total of 58 patients were referred for endoscopic resection of upper gastrointestinal SELs. The median diameter of lesions was 20 mm (range 7-100 mm). Median follow-up time was 5 months (range 0.4-75.7). Forty-eight procedures (83%) were completed successfully leading to en bloc resection in 85% and R0-resection in 63%. Procedure-related AEs occurred in six patients (13%). Severe complications (CD grade 3a) were seen in three patients. The local recurrence rate for (pre)malignant diagnosis was 2%. Additional surgical intervention was needed in seven patients (15%). A total of 32 patients (67%) could be discharged from further surveillance after endoscopic resection. Conclusions Endoscopic resection is a safe and effective treatment for SELs and offers valuable information about undetermined SELs for which repeated sampling attempts have failed to provide adequate tissue for diagnosis.

EUS-guided hydrogel injection to separate pancreatic head carcinoma from duodenum for enhanced radiotherapy: Multi-site feasibility study.

Background and study aims The proximity of a pancreas head tumor to the duodenum often limits delivery of an ablative dose of radiation therapy. This study evaluated the feasibility and safety of using an injectable polyethylene glycol (PEG) hydrogel between the head of the pancreas and duodenum. Patients and methods In a multi-site feasibility cohort study of patients with localized pancreatic cancer, PEG hydrogel was injected under endoscopic ultrasound guidance to temporarily position the duodenum away from the pancreas. Procedure characteristics were recorded, including hydrogel volume and space created. Patients were monitored for adverse events (AEs) and radiotherapy toxicity. Results In all six intent-to-treat patients (four with borderline resectable, two with locally advanced disease), the ability to place and visualize PEG hydrogel and create space between the duodenum and the head of the pancreas was successful. There were no procedure-related AEs resulting in radiotherapy delay. There were no device-related AEs and no reports of pancreatitis. Conclusions PEG hydrogel was successfully placed, created space between the duodenum and the head of the pancreas, and was not associated with major toxicity. Enhancing radiotherapy for pancreatic cancer by using PEG hydrogel to create peri-duodenal space could have beneficial implications for treatment and warrants more exploration.

Comparison of stent patency between EUS-guided hepaticogastrostomy with bridging and endoscopic transpapillary biliary drainage for hilar obstruction.

Background and study aims Endoscopic ultrasound-guided hepaticogastrostomy with bridging between the left and right bile ducts is an alternative to endoscopic transpapillary drainage for malignant hilar biliary obstruction. We aimed to analyze the long-term stent patency of endoscopic ultrasound-guided hepaticogastrostomy with bridging. Patients and methods Patients who underwent endoscopic ultrasound-guided hepaticogastrostomy with bridging between April 2018 and July 2023 were retrospectively analyzed. We retrospectively compared the stent patency of these patients with that of the individuals who underwent endoscopic transpapillary drainage-multi-stenting using unmatched (entire) and propensity score-matched cohorts. Results Endoscopic ultrasound-guided hepaticogastrostomy with bridging had a technical success rate of 90% (18/20). Adverse events were minimal. The number of clinical success cases was 17 and 82 for endoscopic ultrasound-guided hepaticogastrostomy with bridging using metallic stent and endoscopic transpapillary drainage-multi-stenting, respectively. The recurrent biliary obstruction rate was 17.6% and 58.5% for endoscopic ultrasound-guided hepaticogastrostomy with bridging and endoscopic transpapillary drainage-multi-stenting, respectively; the median time to recurrent biliary obstruction (days) was significantly longer for endoscopic ultrasound-guided hepaticogastrostomy with bridging in the entire (not reached vs. 104, P =0.03) and propensity score-matched (183 vs. 79, P =0.05) cohorts. The non-recurrent biliary obstruction rate for endoscopic ultrasound-guided hepaticogastrostomy with bridging was 91.6% at 3 and 6 months and 57% at 12 months. Multivariate analyses revealed that endoscopic ultrasound-guided hepaticogastrostomy with bridging contributed to a lower recurrent biliary obstruction incidence (hazard ratio, 0.31, P =0.05) without significant difference. Conclusions Stent patency was significantly better for endoscopic ultrasound-guided hepaticogastrostomy with bridging. However, future prospective studies are needed.

Endoscopic ultrasonography-based intratumoral and peritumoral machine learning radiomics analyses for distinguishing insulinomas from non-functional pancreatic neuroendocrine tumors.

Objectives: To develop and validate radiomics models utilizing endoscopic ultrasonography (EUS) images to distinguish insulinomas from non-functional pancreatic neuroendocrine tumors (NF-PNETs). Methods: A total of 106 patients, comprising 61 with insulinomas and 45 with NF-PNETs, were included in this study. The patients were randomly assigned to either the training or test cohort. Radiomics features were extracted from both the intratumoral and peritumoral regions, respectively. Six machine learning algorithms were utilized to train intratumoral prediction models, using only the nonzero coefficient features. The researchers identified the most effective intratumoral radiomics model and subsequently employed it to develop peritumoral and combined radiomics models. Finally, a predictive nomogram for insulinomas was constructed and assessed. Results: A total of 107 radiomics features were extracted based on EUS, and only features with nonzero coefficients were retained. Among the six intratumoral radiomics models, the light gradient boosting machine (LightGBM) model demonstrated superior performance. Furthermore, a peritumoral radiomics model was established and evaluated. The combined model, integrating both the intratumoral and peritumoral radiomics features, exhibited a comparable performance in the training cohort (AUC=0.876) and achieved the highest accuracy in predicting outcomes in the test cohorts (AUC=0.835). The Delong test, calibration curves, and decision curve analysis (DCA) were employed to validate these findings. Insulinomas exhibited a significantly smaller diameter compared to NF-PNETs. Finally, the nomogram, incorporating diameter and radiomics signature, was constructed and assessed, which owned superior performance in both the training (AUC=0.929) and test (AUC=0.913) cohorts. Conclusion: A novel and impactful radiomics model and nomogram were developed and validated for the accurate differentiation of NF-PNETs and insulinomas utilizing EUS images.

[Endoscopic ultrasound-guided liver biopsy: current status and prospects].

Liver biopsy is an important means of clinical diagnosis and treatment of liver diseases, but it is not easily accepted by patients because of its invasiveness. The most commonly employed liver biopsy approaches are percutaneous or transjugular. Endoscopic ultrasound-guided liver biopsy (EUS-LB), a newly emerging transjugular technique, has been widely studied and applied in recent years, but its application in China is less common. The EUS-LB has the advantages of high safety and comfort, simultaneous sampling of both liver lobes, and adequate sampling volume; however, it also has the disadvantages of high requirements for hardware, operators, and cost. This article reviews the clinical application of EUS-LB in accordance with pertinent research findings from recent years and discusses its advantages, disadvantages, and implementation feasibility.

Role of endoscopy in eosinophilic esophagitis.

Eosinophilic esophagitis (EoE) is a chronic immune-mediated disease involving inflammation of the esophagus. Endoscopy is essential in the diagnosis and treatment of EoE and shows typical findings, including esophageal edema, rings, exudates, furrows, and stenosis. However, studies involving pediatric and adult patients with EoE suggest that even a normally appearing esophagus can be diagnosed as EoE by endoscopic biopsy. Therefore, in patients with suspected EoE, biopsy samples should be obtained from the esophagus regardless of endoscopic appearance. Moreover, follow-up endoscopies with biopsy after therapy initiation are usually recommended to assess response. Although previous reports of endoscopic ultrasonography findings in patients with EoE have shown diffuse thickening of the esophageal wall, including lamina propria, submucosa, and muscularis propria, its role in EoE remains uncertain and requires further investigation. Endoscopic dilation or bougienage is a safe and effective procedure that can be used in combination with medical and/or dietary elimination therapy in patients with esophageal stricture for the management of dysphagia and to prevent its recurrence.

Multiple intraductal papillary neoplasms of bile duct diagnosed based on endoscopic ultrasonography and peroral cholangioscopy findings.

A man in his 30s was referred to our department for evaluation of bile duct stricture and removal of an intrahepatic bile duct stone. Five years before his presentation, he underwent left hepatectomy for a giant hepatic hemangioma. There were no abnormalities in blood biochemical tests. Magnetic resonance cholangiopancreatography showed one 5 mm oval defect in region B6 and two 8 mm semicircular defects in the hilar bile duct. Endoscopic ultrasound revealed a 3.5 mm hypoechoic focal raised lesion in the hilar bile duct. Oral cholangioscopy revealed his two lesions in the hilar bile duct as white papillary elevations with mucus production. The pathological diagnosis of intraductal papillary neoplasm was determined (low-grade dysplasia, type 1, gastric type). After 1 and a half years, no expansion of the bile duct lesion was observed. Initially, it was thought to be a benign stenosis after liver resection, but based on the results of endoscopic ultrasound, we suspected a tumorous lesion, and we were able to make an accurate diagnosis, including histological type, using transoral cholangioscopy.

Artificial Intelligence-Based Diagnosis of Gastric Mesenchymal Tumors Using Digital Endosonography Image Analysis.

Background/Objectives: Artificial intelligence (AI)-assisted endoscopic ultrasonography (EUS) diagnostic tools have shown excellent performance in diagnosing gastric mesenchymal tumors. This study aimed to assess whether incorporating clinical and endoscopic factors into AI-assisted EUS classification models based on digital image analysis could improve the diagnostic performance of AI-assisted EUS diagnostic tools. Methods: We retrospectively analyzed the data of 464 patients who underwent both EUS and surgical resection of gastric mesenchymal tumors, including 294 gastrointestinal stromal tumors (GISTs), 52 leiomyomas, and 41 schwannomas. AI-assisted classification models for GISTs and non-GIST tumors were developed utilizing clinical and endoscopic factors and digital EUS image analysis. Results: Regarding the baseline EUS classification models, the area under the receiver operating characteristic (AUC) values of the logistic regression, decision tree, random forest, K-nearest neighbor (KNN), and support vector machine (SVM) models were 0.805, 0.673, 0.781, 0.740, and 0.791, respectively. Using the new classification models incorporating clinical and endoscopic factors into the baseline classification models, the AUC values of the logistic regression, decision tree, random forest, KNN, and SVM models increased to 0.853, 0.715, 0.896, 0.825, and 0.794, respectively. In particular, the random forest and KNN models exhibited significant improvement in performance in Delong's test (both p < 0.001). Conclusion: The diagnostic performance of the AI-assisted EUS classification models improved when clinical and endoscopic factors were incorporated. Our results provided direction for developing new AI-assisted EUS models for gastric mesenchymal tumors.

Diagnostic Assessment of Endoscopic Ultrasonography-Fine Needle Aspiration Cytology in the Pancreas: A Comparison between Liquid-Based Preparation and Conventional Smear.

Background and Objectives: This study aimed to elucidate the cytologic characteristics and diagnostic usefulness of endoscopic ultrasonography-fine needle aspiration cytology (EUS-FNAC) by comparing it with liquid-based preparation (LBP) and conventional smear (CS) in pancreas. Methods: The diagnostic categories (I through VII) were classified according to the World Health Organization Reporting System for Pancreaticobiliary Cytopathology. Ten cytologic features, including nuclear and additional features, were evaluated in 53 cases subjected to EUS-FNAC. Nuclear features comprised irregular nuclear contours, nuclear enlargement, hypochromatic nuclei with parachromatin clearing, and nucleoli. Additional cellular features included isolated atypical cells, mucinous cytoplasm, drunken honeycomb architecture, mitosis, necrotic background, and cellularity. A decision tree analysis was conducted to assess diagnostic efficacy. Results: The diagnostic concordance rate between LBP and CS was 49.1% (26 out of 53 cases). No significant differences in nuclear features were observed between categories III (atypical), VI (suspicious for malignancy), and VII (malignant). The decision tree analysis of LBP indicated that cases with moderate or high cellularity and mitosis could be considered diagnostic for those exhibiting nuclear atypia. Furthermore, in CS, mitosis, isolated atypical cells, and necrotic background exerted a more significant impact on the diagnosis of EUS-FNAC. Conclusions: Significant parameters for interpreting EUS-FNAC may differ between LBP and CS. While nuclear atypia did not influence the diagnosis of categories III, VI, and VII, other cytopathologic features, such as cellularity, mitosis, and necrotic background, may present challenges in diagnosing EUS-FNAC.

Endoscopic ultrasound-guided fine-needle aspiration in the diagnostic value of focal liver lesions: A systematic analysis of 88 cases (with video).

BACKGROUND AND AIMS: Endoscopic ultrasonography-guided fine-needle aspiration (EUS-FNA) is an important diagnostic tool for suspected parenchymal lesions in the gastrointestinal tract and adjacent organs. Our study aimed to evaluate the safety and effectiveness of EUS-FNA in focal liver lesions (FLLs). METHOD: Data from 88 patients diagnosed with FLLs by imaging who underwent EUS-FNA from 1 January 2017 to 31 August 2022 were reviewed in our retrospective study at the Second Affiliated Hospital of Soochow University and Ruijin Hospital of the School of Medicine of Shanghai Jiao Tong University. The EUS-FNA biopsy results were compared with the final diagnosis to evaluate diagnostic value. The relevant factors were analysed to determine their influence on EUS-FNA biopsy results. RESULTS: The 88 patients analysed in this study resulted in a final diagnosis of 86 malignant and two benign cases. The overall diagnostic accuracy of EUS-FNA in FLLs was 93.18 % (82/88; 95 % Confidence Interval [CI], 87.9-98.5), with a sensitivity, specificity, positive predictive value, and negative predictive value of 93.02 % (80/86; 95 %CI, 87.6-98.4), 100 % (2/2; 95 %CI, 100-100), 100 % (80/80; 95 %CI, 100-100), and 25 % (2/8; 95 %CI, -5-55.0), respectively. The parameters related to lesion and procedure were not significantly different between these two groups (p > 0.05). The number of puncture needles in the groups showed a statistically significant difference between multiple and single punctures (p = 0.001). CONCLUSION: Our data revealed that EUS-FNA is a safe and reliable diagnostic method for FLLs that shows high accuracy.

SSM-Net: Semi-supervised multi-task network for joint lesion segmentation and classification from pancreatic EUS images.

Pancreatic cancer does not show specific symptoms, which makes the diagnosis of early stages difficult with established image-based screening methods and therefore has the worst prognosis among all cancers. Although endoscopic ultrasonography (EUS) has a key role in diagnostic algorithms for pancreatic diseases, B-mode imaging of the pancreas can be affected by confounders such as chronic pancreatitis, which can make both pancreatic lesion segmentation and classification laborious and highly specialized. To address these challenges, this work proposes a semi-supervised multi-task network (SSM-Net) to leverage unlabeled and labeled EUS images for joint pancreatic lesion classification and segmentation. Specifically, we first devise a saliency-aware representation learning module (SRLM) on a large number of unlabeled images to train a feature extraction encoder network for labeled images by computing a contrastive loss with a semantic saliency map, which is obtained by our spectral residual module (SRM). Moreover, for labeled EUS images, we devise channel attention blocks (CABs) to refine the features extracted from the pre-trained encoder on unlabeled images for segmenting lesions, and then devise a merged global attention module (MGAM) and a feature similarity loss (FSL) for obtaining a lesion classification result. We collect a large-scale EUS-based pancreas image dataset (LS-EUSPI) consisting of 9,555 pathologically proven labeled EUS images (499 patients from four categories) and 15,500 unlabeled EUS images. Experimental results on the LS-EUSPI dataset and a public thyroid gland lesion dataset show that our SSM-Net clearly outperforms state-of-the-art methods.