OBJECTIVES: In recent years, the incidence and detection rate of pancreatic cystic lesions (PCLs) have increased significantly. Endoscopic ultrasound (EUS) plays an indispensable role in the diagnosis and differential diagnosis of PCLs. However, evidence comparing the diagnostic performance of EUS-guided fine-needle aspiration (EUS-FNA) and fine-needle biopsy (FNB) remains limited. This study aims to compare the diagnostic yield, adequacy of tissue acquisition, and safety between EUS-FNA and EUS-FNB in evaluating PCLs to inform clinical practice. METHODS: A retrospective review was conducted on patients with PCLs who underwent either EUS-FNA or EUS-FNB between January 2014 and August 2021. The diagnostic yield, tissue acquisition adequacy, and incidence of adverse events were compared between the 2 groups. RESULTS: A total of 90 patients with PCLs were included (52 in the FNA group and 38 in the FNB group). The diagnostic yield was similar between the FNA and FNB groups (94.2% vs 94.7%, P>0.05). The adequacy of tissue acquisition was 71.2% in the FNA group and 81.6% in the FNB group (P>0.05). No statistically significant difference was observed in the incidence of adverse events between the 2 groups (P>0.05). CONCLUSIONS Both EUS-FNA and EUS-FNB demonstrate equally high diagnostic yields and tissue adequacy in PCLs, with excellent safety profiles. Both methods are safe and effective diagnostic tools for evaluating PCLs.
Humans ; Endoscopic Ultrasound-Guided Fine Needle Aspiration/adverse effects* ; Retrospective Studies ; Female ; Male ; Pancreatic Cyst/diagnostic imaging* ; Middle Aged ; Biopsy, Fine-Needle/adverse effects* ; Aged ; Pancreatic Neoplasms/diagnosis* ; Adult ; Endosonography/methods* ; Pancreas/pathology* ; Diagnosis, Differential

Accurate segmentation of whole slide images is of great significance for the diagnosis of pancreatic cancer. However, developing an automatic model is challenging due to the complex content, limited samples, and high sample heterogeneity of pathological images. This paper presented a multi-tissue segmentation model for whole slide images of pancreatic cancer. We introduced an attention mechanism in building blocks, and designed a multi-task learning framework as well as proper auxiliary tasks to enhance model performance. The model was trained and tested with the pancreatic cancer pathological image dataset from Shanghai Changhai Hospital. And the data of TCGA, as an external independent validation cohort, was used for external validation. The F1 scores of the model exceeded 0.97 and 0.92 in the internal dataset and external dataset, respectively. Moreover, the generalization performance was also better than the baseline method significantly. These results demonstrate that the proposed model can accurately segment eight kinds of tissue regions in whole slide images of pancreatic cancer, which can provide reliable basis for clinical diagnosis.
Humans ; China ; Pancreatic Neoplasms/diagnostic imaging* ; Learning

BACKGROUND: Patients with mass-forming pancreatitis (MFP) or pancreatic ductal adenocarcinoma (PDAC) presented similar clinical symptoms, but required different treatment approaches and had different survival outcomes. This meta-analysis aimed to compare the diagnostic performance of contrast-enhanced ultrasound (CEUS) and contrast-enhanced computed tomography (CECT) in differentiating MFP from PDAC. METHODS: A literature search was performed in the PubMed, EMBASE (Ovid), Cochrane Library (CENTRAL), China National Knowledge Infrastructure (CNKI), Weipu (VIP), and WanFang databases to identify original studies published from inception to August 20, 2021. Studies reporting the diagnostic performances of CEUS and CECT for differentiating MFP from PDAC were included. The meta-analysis was performed with Stata 15.0 software. The outcomes included the pooled sensitivity, specificity, positive likelihood ratio (+LR), negative likelihood ratio (-LR), diagnostic odds ratio (DOR), and summary receiver operating characteristic (SROC) curves of CEUS and CECT. Meta-regression was conducted to investigate heterogeneity. Bayesian network meta-analysis was conducted to indirectly compare the overall diagnostic performance. RESULTS: Twenty-six studies with 2115 pancreatic masses were included. The pooled sensitivity and specificity of CEUS for MFP were 82% (95% confidence interval [CI], 73%-88%; I2  = 0.00%) and 95% (95% CI, 90%-97%; I2  = 63.44%), respectively; the overall +LR, -LR, and DOR values were 15.12 (95% CI, 7.61-30.01), 0.19 (95% CI, 0.13-0.29), and 78.91 (95% CI, 30.94-201.27), respectively; and the area under the SROC curve (AUC) was 0.90 (95% CI, 0.87-92). However, the overall sensitivity and specificity of CECT were 81% (95% CI, 75-85%; I2  = 66.37%) and 94% (95% CI, 90-96%; I2  = 74.87%); the overall +LR, -LR, and DOR values were 12.91 (95% CI, 7.86-21.20), 0.21 (95% CI, 0.16-0.27), and 62.53 (95% CI, 34.45-113.51), respectively; and, the SROC AUC was 0.92 (95% CI, 0.90-0.94). The overall diagnostic accuracy of CEUS was comparable to that of CECT for the differential diagnosis of MFP and PDAC (relative DOR 1.26, 95% CI [0.42-3.83], P  > 0.05). CONCLUSIONS CEUS and CECT have comparable diagnostic performance for differentiating MFP from PDAC, and should be considered as mutually complementary diagnostic tools for suspected focal pancreatic lesions.
Humans ; Contrast Media ; Bayes Theorem ; Tomography, X-Ray Computed/methods* ; Pancreatic Neoplasms/diagnostic imaging* ; Carcinoma, Pancreatic Ductal/diagnostic imaging* ; Sensitivity and Specificity ; Pancreatitis/diagnostic imaging* ; Ultrasonography/methods*

OBJECTIVE: To propose a non-contrast CT-based algorithm for automated and accurate detection of pancreatic lesions at a low cost. METHODS: With Faster RCNN as the benchmark model, an advanced Faster RCNN (aFaster RCNN) model for pancreatic lesions detection based on plain CT was constructed. The model uses the residual connection network Resnet50 as the feature extraction module to extract the deep image features of pancreatic lesions. According to the morphology of pancreatic lesions, 9 anchor frame sizes were redesigned to construct the RPN module. A new Bounding Box regression loss function was proposed to constrain the training process of RPN module regression subnetwork by comprehensively considering the constraints of the lesion shape and anatomical structure. Finally, a detection frame was generated using the detector in the second stage. The data from a total of 728 cases of pancreatic diseases from 4 clinical centers in China were used for training (518 cases, 71.15%) and testing (210 cases, 28.85%) of the model. The performance of aFaster RCNN was verified through ablation experiments and comparison experiments with 3 classical target detection models SSD, YOLO and CenterNet. RESULTS: The aFaster RCNN model for pancreatic lesion detection achieved recall rates of 73.64% at the image level and 92.38% at the patient level, with an average precision of 45.29% and 53.80% at the image and patient levels, respectively, which were higher than those of the 3 models for comparison. CONCLUSION The proposed method can effectively extract the imaging features of pancreatic lesions from non-contrast CT images to detect the pancreatic lesions.
Humans ; Pancreas/diagnostic imaging* ; Algorithms ; China ; Pancreatic Neoplasms/diagnostic imaging* ; Tomography, X-Ray Computed

Objective To determine the appropriate averaging strategy for pancreatic perfusion datasets to create images for routine reading of insulinoma.Methods Thirty-nine patients undergoing pancreatic perfusion CT in Peking Union Medical College Hospital and diagnosed as insulinoma by pathology were enrolled in this retrospective study.The time-density curve of abdominal aorta calculated by software dynamic angio was used to decide the timings for averaging.Five strategies,by averaging 3,5,7,9 and 11 dynamic scans in perfusion,all including peak enhancement of the abdominal aorta,were investigated in the study.The image noise,pancreas signal-to-noise ratio(SNR),lesion contrast and lesion contrast-to-noise ratio(CNR)were recorded and compared.Besides,overall image quality and insulinoma depiction were also compared.ANOVA and Friedman's test were performed.Results The image noise decreased and the SNR of pancreas increased with the increase in averaging time points(all P<0.001).The lesion contrast(69.81±41.35)averaged from 5 scans showed no significant difference compared with that(72.77±45.25)averaged from 3 scans(P=0.103),both of which were higher than that in other groups(all P≤0.001).The lesion CNRs of the last four groups showed no significant difference(all P>0.99)and were higher than that of the first group(all P<0.05).There was no significant difference in overall image quality among the 5 groups(P=0.977).Conclusions Image averaged from 5 scans showed moderate image noise,pancreas SNR and relatively high lesion contrast and lesion CNR.Therefore,it is advised to be used in image averaging to detect insulinoma.
Contrast Media ; Humans ; Insulinoma/diagnostic imaging* ; Pancreas/diagnostic imaging* ; Pancreatic Neoplasms/diagnostic imaging* ; Perfusion ; Radiographic Image Interpretation, Computer-Assisted ; Reading ; Retrospective Studies ; Signal-To-Noise Ratio

To prospectively evaluate the correlation between intravoxel incoherent motion (IVIM)-derived parameters and CT perfusion parameters as well as the pathological grade in insulinoma. A total of 55 patients with suspected insulinoma undergoing IVIM and CT perfusion scans were prospectively enrolled. The images were post-processed to obtain IVIM parameters including apparent diffusion coefficient (ADC),diffusion (D),perfusion correlated diffusion (D*),and f,and CT perfusion parameters including blood flow (BF),blood volume (BV),and permeability (PM). The pathological specimens were stained to obtain pathological parameters including the grading,ki-67 index,and the mitotic count. The IVIM derived parameters of normal pancreas including head,body,and tail as well as that of the pancreatic insulinoma were compared. The correlation between IVIM parameters and CT perfusion parameters as well as the pathological parameters was analyzed. ADC and D values of pancreatic tail were significantly lower than those of the pancreatic head and neck (all <0.001). There were significant differences in all IVIM parameters between insulinoma and normal pancreas (all <0.001). The ADC and f value of the normal pancreas was positively correlated with BF (=0.437,=0.003;=0.357,=0.010). There is no correlation between the remaining IVIM parameters and the CT perfusion parameters as well as between IVIM parameters and pathological parameters (all >0.05). IVIM parameters differ at different anatomical parts of normal pancreas. IVIM parameters can distinguish normal pancreatic parenchyma from insulinoma. The ADC value is weakly correlated with BF.
Diffusion Magnetic Resonance Imaging ; Humans ; Insulinoma ; diagnostic imaging ; Motion ; Pancreatic Neoplasms ; diagnostic imaging ; Reproducibility of Results ; Tomography, X-Ray Computed

A 70-year-old male presenting with a mass in the right inguinal area was treated with surgery, and was diagnosed pathologically as spermatic cord metastasis of pancreatic cancer. He was given systemic chemotherapy. Unfortunately, he died of ascites and cachexia three months later.
Aged ; Fatal Outcome ; Genital Diseases, Male/surgery* ; Humans ; Male ; Pancreatic Neoplasms/diagnostic imaging* ; Spermatic Cord/surgery* ; Tomography, X-Ray Computed/methods*

Objective To evaluate the value of texture features derived from intravoxel incoherent motion (IVIM) parameters for differentiating pancreatic neuroendocrine tumor (pNET) from pancreatic adenocarcinoma (PAC).Methods Eighteen patients with pNET and 32 patients with PAC were retrospectively enrolled in this study. All patients underwent diffusion-weighted imaging with 10 b values used (from 0 to 800 s/mm ). Based on IVIM model, perfusion-related parameters including perfusion fraction (f), fast component of diffusion (D) and true diffusion parameter slow component of diffusion (D) were calculated on a voxel-by-voxel basis and reorganized into gray-encoded parametric maps. The mean value of each IVIM parameter and texture features [Angular Second Moment (ASM), Inverse Difference Moment (IDM), Correlation, Contrast and Entropy] values of IVIM parameters were measured. Independent sample t-test or Mann-Whitney U test were performed for the between-group comparison of quantitative data. Regression model was established by using binary logistic regression analysis, and receiver operating characteristic (ROC) curve was plotted to evaluate the diagnostic efficiency.Results The mean f value of the pNET group were significantly higher than that of the PAC group (27.0% vs. 19.0%, P = 0.001), while the mean values of D and D showed no significant differences between the two groups. All texture features (ASM, IDM, Correlation, Contrast and Entropy) of each IVIM parameter showed significant differences between the pNET and PAC groups (P=0.000-0.043). Binary logistic regression analysis showed that texture ASM of D and texture Correlation of D were considered as the specific imaging variables for the differential diagnosis of pNET and PAC. ROC analysis revealed that multiple texture features presented better diagnostic performance than IVIM parameters (AUC 0.849-0.899 vs. 0.526-0.776), and texture ASM of D combined with Correlation of D in the model of logistic regression had largest area under ROC curve for distinguishing pNET from PAC (AUC 0.934, cutoff 0.378, sensitivity 0.889, specificity 0.854).Conclusions Texture analysis of IVIM parameters could be an effective and noninvasive tool to differentiate pNET from PAC.
Adenocarcinoma ; diagnostic imaging ; Adult ; Aged ; Algorithms ; Diagnosis, Differential ; Diffusion Magnetic Resonance Imaging ; Female ; Humans ; Male ; Middle Aged ; Motion ; Pancreatic Neoplasms ; diagnostic imaging ; Retrospective Studies

Autoimmune pancreatitis (AIP) is a unique subtype of chronic pancreatitis, which shares many clinical presentations with pancreatic ductal adenocarcinoma (PDA). The misdiagnosis of AIP often leads to unnecessary pancreatic resection. F-FDG positron emission tomography/ computed tomography (PET/CT) could provide comprehensive information on the morphology, density, and functional metabolism of the pancreas at the same time. It has been proved to be a promising modality for noninvasive differentiation between AIP and PDA. However, there is a lack of clinical analysis of PET/CT image texture features. Difficulty still remains in differentiating AIP and PDA based on commonly used diagnostic methods. Therefore, this paper studied the differentiation of AIP and PDA based on multi-modality texture features. We utilized multiple feature extraction algorithms to extract the texture features from CT and PET images at first. Then, the Fisher criterion and sequence forward floating selection algorithm (SFFS) combined with support vector machine (SVM) was employed to select the optimal multi-modality feature subset. Finally, the SVM classifier was used to differentiate AIP from PDA. The results prove that texture analysis of lesions helps to achieve accurate differentiation of AIP and PDA.
Adenocarcinoma ; diagnostic imaging ; Algorithms ; Autoimmune Diseases ; diagnostic imaging ; Diagnosis, Differential ; Fluorodeoxyglucose F18 ; Humans ; Pancreatic Neoplasms ; diagnostic imaging ; Pancreatitis ; diagnostic imaging ; Positron Emission Tomography Computed Tomography ; Support Vector Machine

Objective To characterize the CT perfusion parameters of focal pancreatic lesions including pancreatic cancers (PACs) and pancreatic neuroendocrine tumors (pNETs),estimate the confirmity and fungibility of parameters obtained from Deconvolution and Maximum slope+Patlak.Methods From December 2015 to November 2016,22 patients with PACs and 22 patients with pNETs(37 lesions confirmed by surgery and biopsy)underwent preoperative whole-pancreas CT perfusion in our center. The volume perfusion CT of the entire pancreas was performed at 80 kV and 100 mA,using 28 consecutive volume measurements and intravenous injection of 45 ml of iodinated contrast and saline at a flow rate of 5 ml/s. One experienced radiologists measured and recorded the CT perfusion parameters on Siemens post-processing workstation using two mathematical methods:Maximum slope+Patlak analysis versus Deconvolution method.ResultsWilcoxon matched-pairs test revealed significant difference between both pairs of the perfusion measurements by the two methods,PACs(BFM vs. BFD,Z=-3.263,P=0.001;BVD vs. BVP,Z=-3.978,P=0.000); pNETs(BFM vs. BFD,Z=-5.212,P=0.000;BVD vs. BVP,Z=-2.633,P=0.008). Spearman's correlation coefficient showed both pairs of perfusion measurements significantly correlated with each other in PACs (BFM vs. BFD,r=0.845,P=0.000;BVD vs. BVP,r=0.964,P=0.000) and pNETs(BFM vs. BFD,r=0.759,P=0.000),BVD vs. BVP,r=0.683,P=0.000). Geometric mean BFM/BFD ratio in PACs was 0.77 (range:0.61-0.99),while geometric mean BVD/BVP ratio was 1.42 (range:1.13-1.79),within 95% limits of agreement. Geometric mean BFM/BFD ratio in pNETs was 0.66 (range:0.51-0.86),while geometric mean BVD/BVP ratio was 1.15 (range:0.88-1.50),within 95% limits of agreement. Conclusion sSignificantly different CT perfusion values of blood flow and blood volume were obtained by Deconvolution-based and Maximum slope+Patlak-based algorithms in the pNETs and PACs. They correlated significantly with each other. Two perfusion-measuring algorithms are interchangeable because the ranges of the conversion factors are narrow.
Algorithms ; Blood Volume ; Contrast Media ; Humans ; Pancreas ; diagnostic imaging ; pathology ; Pancreatic Neoplasms ; diagnostic imaging ; Reproducibility of Results ; Sensitivity and Specificity ; Tomography, X-Ray Computed ; methods