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

OBJECTIVETo investigate the value of contrast-enhanced ultrasound in the differential diagnosis of solid focal lesions of pancreas (s-FLPs).

METHODSWe retrospectively analyzed the clinical data of 56 s-FLPs examined with contrast agent combined with low mechanical indicators contrast-enhanced ultrasound.

RESULTSThe surrounding pancreas parenchyma enhancement time, lesion enhancement time, and peak enhancement time between different groups of s-FLPs had no significant differences (P > 0.05), while the beginning to peak enhancement time showed significant difference (P < 0.05). When using the enhancement speed as a diagnostic indicator to differentiate pancreatic carcinoma from tumor-like pancreatitis, the sensitivity, specificity, and accuracy were 90.5%, 71.4%, and 85.7% for pancreatic carcinoma and 75.0%, 91.7%, and 89.3% for tumor-like pancreatitis. When using the enhancement pattern as a diagnosis indicator to differentiate pancreatic carcinoma from tumor-like pancreatitis, the sensitivity, specificity, and accuracy were 85.7%, 78.6%, and 83.9% for pancreatic carcinoma and 75.0%, 100%, and 94.6% for tumor-like pancreatitis. When different indicators were combined, enhancement pattern and enhancement speed showed the best diagnostic results; however, the Youden index was not improved.

CONCLUSIONSDifferent s-FLPs show different enhancement findings on contrast-enhanced ultrasound. The enhancement pattern and enhancement speed are the most useful diagnostic indicators.

Diagnosis, Differential ; Humans ; Pancreas ; diagnostic imaging ; Pancreatic Neoplasms ; diagnostic imaging ; Pancreatitis ; diagnostic imaging ; Retrospective Studies ; Ultrasonography

Detection of cystic lesions in the pancreas has increased because of the widespread use of high-resolution diagnostic imaging techniques. Therefore, cystic lesions of the pancreas constitute an increasingly important category with a challenging differential diagnosis. Squamoid cyst of pancreatic ducts is a recently recognized type of cystic lesion in the pancreas in which cystically dilated ducts are lined by non-keratinized squamous epithelium. Although it is clinically known as benign cystic lesion, we experienced its malignant behavior and report here with review of the international literatures.
Diagnosis, Differential ; Diagnostic Imaging ; Epithelium ; Pancreas ; Pancreatic Cyst ; Pancreatic Ducts* ; Pancreatic Neoplasms

Objective To compare the sensitivity of multislice spiral CT dual phase and somatosatatin receptor scintigraphy (SRS) in the diagnosis of pancreas nuroendocrine tumors (pNET). Methods Totally 28 patients with pathologically confirmed pNET recieved both CT dual phase contrast and SRS and the results were compared. Results Of these 28 pNET patients,26 (92.8%) were accurately diagnosed by CT dual-phase scan and 20 (71.4%) by SRS (P＝0.031).In the functioning pNET cases,the diagnosis sensitivity of CT dual phase scan and SRS was 94.1% (16/17)and 58.8% (10/17)(P＝0.218). In the non-functioning pNET cases,the sensitivity was 90.9% (10/11) and 90.9% (10/11) (P＝0.740).Diagnostic sensitivity of CT dual phase scan and SRS for pNET without metastasis was 90.4% (19/21) and 57.1% (12/21) (P＝0.125).The sensitivity for pNET with metastasis was 100%(7/7)and 100% (7/7). Corresponding to the pathological grading,the diagnostic sensitivity of CT dual phase scanning and SRS was 84.6% (11/13) and 53.8% (7/13) for G1,100% (12/12) and 83.3% (10/12) for G2,and 100% (3/3) and 100% (3/3) for G3. The diagnostic sensitivity of CT dual phase scan and SRS for pNET with diameter less than or equal to 2.0 cm was 94.7% (18/19) and 52.6% (10/19) (P＝0.008). For pNET with diameter more than 2.0 cm,the sensitivity was 92.8% (13/14) and 100% (14/14). Conclusions Compared with SRS,dual phase CT scan is more sensitive in diagnosing pNET,especially for those in lower pathological stages. For lesions sized less than or equal to 2.0 cm,SRS should be combined with other imaging examinations to minimize false negative results.
Humans ; Neuroendocrine Tumors ; diagnostic imaging ; Pancreatic Neoplasms ; diagnostic imaging ; Radionuclide Imaging ; Sensitivity and Specificity ; Tomography, Spiral Computed

OBJECTIVETo evaluate the clinicopathological and CT features of mucinous cystic neoplasms (MCNs) of the pancreas and analyze the correlative risk factors for malignant pancreatic mucinous cystic neoplasms.

METHODSNinety-eight patients who underwent curative resection for mucinous cystic neoplasms of the pancreas at PLA General Hospital from April 1994 to January 2013 were included in this study. All clinicopathological data available were retrospectively analyzed. All patients were divided into benign tumors + premalignant lesion group (70 patients) and malignant tumor group (28 patients). Clinicopathological and CT features of the mucinous cystic neoplasms of the pancreas and risk factors of malignant pancreatic mucinous cystic neoplasms were analyzed.

RESULTSMucinous cystic neoplasms were seen mostly in perimenopausal women (71.4%, 70/98 cases,), and 51.0% (50/98 cases) of the patients had obvious clinical signs, mostly non-specific abdominal pain, but jaundice was present only in cases of malignant mucinous cystic neoplasms. Benign mucinous cystic neoplasms were mostly located in the distal pancreas (74.3%) and characterized with septa and thin cystic wall, while more malignant mucinous cystic neoplasms were located at the proximal pancreas (57.1%) and characterized with thick cystic wall and solid components. Univariate analysis showed that findings associated with malignancy gender, age ≥ 60, presence of symptoms, jaundice, weight loss, tumor location, margin, wall thickness, solid components and dilation of the main pancreatic duct were significantly correlated with malignant tumor development (P < 0.05 for all). The results of multiple logistic regression analysis showed that thick wall and solid components were independent prognostic factors for malignancy (OR = 31.417 and 34.976, P < 0.05 for both).

CONCLUSIONSGender, age ≥ 60, presence of symptoms, jaundice, weight loss, tumor location, margin, wall thickness, solid components and dilation of the main pancreatic duct are important diagnostic indices of malignant mucinous cystic tumors of the pancreas, while thick wall and solid components are independent risk factors of malignant pancreatic mucinous cystic neoplasms.

Adenocarcinoma, Mucinous ; diagnostic imaging ; Aged ; Female ; Humans ; Neoplasms, Glandular and Epithelial ; diagnostic imaging ; Pancreatic Neoplasms ; diagnostic imaging ; Radiography ; Retrospective Studies

Adult ; Female ; Humans ; Pancreatic Neoplasms ; diagnosis ; diagnostic imaging ; Radiography ; Tuberculosis, Lymph Node ; diagnosis ; diagnostic imaging

OBJECTIVEThis pictorial review aimed to summarize the most possible differential diagnosis of pancreatic islet cell tumor (PICT).

DATA SOURCESData used in this review were mainly from Medline and Pubmed in English. And all clinical images in this review were from Department of Radiology, Peking Union Medical College Hospital, Beijing, China.

STUDY SELECTIONCases of pancreatic cystadenoma, solid pseudo-papillary tumor of the pancreas, pancreatic metastasis, pancreatic adenocarcinoma, para-pancreatic neuroendocrine tumors, Castleman disease, gastrointestinal stromal tumor, splenic artery aneurysm and accessory spleen were selected in this pictorial review for differential diagnosis of PICT.

RESULTSCareful analysis of imaging features and correlation with the clinical manifestations may allow a more specific diagnosis. It is also important that the radiologist is familiar with the anatomic variants and disease entities which mimic pancreatic islet cell tumor in order to avoid an improper treatment protocol.

CONCLUSIONSMany congenital anatomic variants or other pancreatic and peri-pancreatic diseases may mimic MDCT appearance of pancreatic islet cell tumor. Radiological, clinical and pathological characteristics should be considered for the final diagnosis.

Humans ; Neuroendocrine Tumors ; diagnosis ; diagnostic imaging ; Pancreatic Neoplasms ; diagnosis ; diagnostic imaging ; Radiography

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

Although the tail of the pancreas is the second most common site of an accessory spleen, intrapancreatic accessory spleen (IPAS) has rarely been noted radiologically. However, as the imaging techniques have recently advanced, IPAS will be more frequently detected as an incidental pancreatic nodule on CT or MRI. Because accessory spleens usually pose no clinical problems, it is important to characterize accessory spleens as noninvasively as possible. An IPAS has similar characteristics to those of the spleen on the precontrast and contrast-enhanced images of all the imaging modalities. In particular, inhomogeneous enhancement of an IPAS in its early phases may be a diagnostic clue. Superparamagnetic iron oxide (SPIO)-enhanced MRI and Levovist-enhanced US, and the mechanisms of which are theoretically similar to that of Tc-99m scintigraphy, can be used as alternative tools to confirm the diagnosis of IPAS. An IPAS shows a significant signal drop similar to the spleen on the SPIO-enhanced T2 or T2*-weighted imaging and prolonged enhancement on the delayed hepatosplenic phase of contrast-enhanced US. We review and illustrate the differential points between IPAS and hypervascular pancreatic tumors in this manuscript.
Choristoma/*diagnosis ; Diagnosis, Differential ; *Diagnostic Imaging ; Humans ; Pancreatic Diseases/*diagnosis ; Pancreatic Neoplasms/diagnosis ; *Spleen

OBJECTIVETo analyze the clinical and ultrasonographic imaging features of intraductal papillary mucinous neoplasm (IPMN) of the pancreas.

METHODSTwelve patients with IPMN underwent surgery between May 2005 and December 2008, including 4 (33.3%) with adenoma and 8 (66.7%) with adenocarcinoma. IPMN was classified preoperatively into 3 types based on sonographic findings of different sites: main duct, branch duct, and combined type. All clinical presentations and ultrasonographic findings of those patients were reviewed and the correlation between ultrasonographic findings and histopathological results was analyzed.

RESULTSThere were 9 men and 3 women with a mean age of 60.1 +/- 9.6 years (range, 32-73). Of all the 12 patients with IPMN, 9 (75.0%) had experienced some symptoms of epigastric discomfort and/or pain as well as backache; 7 cases were with medical history of acute pancreatitis, 5 cases with diabetes, 4 cases with elevated CA19-9, and 2 cases with steatorrhea. All lesions of IPMN have been revealed by transabdominal ultrasonography. The mean diameters of the lesions were 1.4 +/- 0.8 cm (range, 0.5-2.0) and 6.3 +/- 6.0 cm (range, 2.0-20.0) in adenomas and adenocarcinomas, respectively. And the mean diameters of the main duct in adenomas and adenocarcinomas were 1.0 +/- 0.8 cm and 1.6 +/- 1.0 cm, respectively. Among the 4 adnomas, 3 (75.0%) cases were classified as branch type based on sonographic findings, and 2 were demonstrated as mural nodules in which no color signals was detected. Among the 8 adenocarcinomas, 5 (62.5%) cases were classified as main duct type, and 3 (37.5%) as combined type. In 7 of the 8 adenocarcinomas, mural nodules were detected within the dilated ducts or cysts of the lesions in which color flow signals were detected.

CONCLUSIONSTransabdominal ultrasonography can reveal the pancreatic cystic lesions of IPMN as well as dilated pancreatic ducts. Some characteristics should be noticed as suggesting the possibility of malignancy: clinical symptoms of pancreatic insufficiency, large tumor size, and mural nodules with color Doppler flow signals. Transabdominal ultrasonography could be a useful tool to help diagnose and make appropriate management of IPMN.

Adenocarcinoma, Mucinous ; diagnostic imaging ; Adult ; Aged ; Carcinoma, Pancreatic Ductal ; diagnostic imaging ; Carcinoma, Papillary ; diagnostic imaging ; Female ; Humans ; Male ; Middle Aged ; Pancreatic Cyst ; diagnostic imaging ; Pancreatic Neoplasms ; diagnostic imaging ; Ultrasonography