New criteria to differentiate between mucinous cystic neoplasm and serous cystic neoplasm in pancreas by endoscopic ultrasound: A preliminarily confirmed outcome of 41 patients

Artificial intelligence for the detection of pancreatic lesions

PURPOSE

Pancreatic cancer is one of the most lethal neoplasms among common cancers worldwide, and PCLs are well-known precursors of this type of cancer. Artificial intelligence (AI) could help to improve and speed up the detection and classification of pancreatic lesions. The aim of this review is to summarize the articles addressing the diagnostic yield of artificial intelligence applied to medical imaging (computed tomography [CT] and/or magnetic resonance [MR]) for the detection of pancreatic cancer and pancreatic cystic lesions.

METHODS

We performed a comprehensive literature search using PubMed, EMBASE, and Scopus (from January 2010 to April 2021) to identify full articles evaluating the diagnostic accuracy of AI-based methods processing CT or MR images to detect pancreatic ductal adenocarcinoma (PDAC) or pancreatic cystic lesions (PCLs).

RESULTS

We found 20 studies meeting our inclusion criteria. Most of the AI-based systems used were convolutional neural networks. Ten studies addressed the use of AI to detect PDAC, eight studies aimed to detect and classify PCLs, and 4 aimed to predict the presence of high-grade dysplasia or cancer.

CONCLUSION

AI techniques have shown to be a promising tool which is expected to be helpful for most radiologists' tasks. However, methodologic concerns must be addressed, and prospective clinical studies should be carried out before implementation in clinical practice.

Deep Learning-Based Differentiation between Mucinous Cystic Neoplasm and Serous Cystic Neoplasm in the Pancreas Using Endoscopic Ultrasonography

Mucinous cystic neoplasms (MCN) and serous cystic neoplasms (SCN) account for a large portion of solitary pancreatic cystic neoplasms (PCN). In this study we implemented a convolutional neural network (CNN) model using ResNet50 to differentiate between MCN and SCN. The training data were collected retrospectively from 59 MCN and 49 SCN patients from two different hospitals. Data augmentation was used to enhance the size and quality of training datasets. Fine-tuning training approaches were utilized by adopting the pre-trained model from transfer learning while training selected layers. Testing of the network was conducted by varying the endoscopic ultrasonography (EUS) image sizes and positions to evaluate the network performance for differentiation. The proposed network model achieved up to 82.75% accuracy and a 0.88 (95% CI: 0.817–0.930) area under curve (AUC) score. The performance of the implemented deep learning networks in decision-making using only EUS images is comparable to that of traditional manual decision-making using EUS images along with supporting clinical information. Gradient-weighted class activation mapping (Grad-CAM) confirmed that the network model learned the features from the cyst region accurately. This study proves the feasibility of diagnosing MCN and SCN using a deep learning network model. Further improvement using more datasets is needed.

Surveillance and management for serous cystic neoplasms of the pancreas based on total hazards-a multi-center retrospective study from China

Background

Serous cystic neoplasms (SCN) rarely have malignant potential, so accurate diagnosis of SCN is crucial for proper clinical management, especially to avoid unnecessary surgeries. However, the misdiagnosis of other pancreatic cystic neoplasm instead of SCN may highly increase the risk of malignancy in patients who receive no surgery.

Methods

Data from a total of 678 patients with pathologically confirmed to have SCN at sixteen institutions in China from January 1^st, 2006 to December 31^st, 2016 were retrieved to evaluate the malignancy risk of SCN.

Results

Among the 678 patients confirmed to have SCN with postoperative pathologic analysis, 649 patients (95.7%) had only one lesion and the average maximum diameter was 3.8±2.47 cm. Four patients were pathologically verified as having serous cystadenocarcinoma, so the SCN actual malignancy rate was 0.6%, while the mortality due to pancreatic surgery in these high-volume centers was nearly 0.2-2%. However, among the 99 SCN patients based on preoperative radiology, three were confirmed to have intraductal papillary mucinous neoplasms (IPMN), nine as mucinous cystic neoplasms (MCN), and four as solid pseudopapillary tumors (SPT) after postoperative pathological analysis. Thus, the total theoretical malignancy rate resulting from preoperative misdiagnosis was elevated to approximately 2.9%, higher than the risk of perioperative mortality.

Conclusions

When SCN can't be accurately distinguished from cystic tumors of pancreas, the malignant risk of cystic tumors may be higher than perioperative risk. However, if it can be diagnosed as SCN accurately, surgery can be avoided as well.

Endoscopic ultrasonography for the evaluation of pancreatic cystic neoplasms

Endoscopic ultrasonography (EUS) is a modality with high spatial resolution that enables comprehensive observation of the entire pancreas and plays an important role in the diagnosis of pancreatic lesions. Recent advances in diagnostic imaging methods such as ultrasound, computed tomography, and magnetic resonance imaging have increased the incidental detection of pancreatic cystic lesions (PCLs). EUS has been recognized as an essential diagnostic method for the detection and evaluation of PCLs. EUS has two important roles: as a detailed (high-resolution) imaging diagnostic method and as an approach for collecting cyst fluid content by EUS-guided fine needle aspiration for pathological diagnosis or biomarker evaluation. Furthermore, in recent years, the usefulness of contrast-enhanced EUS for the differential diagnosis of PCLs or evaluation of grade of malignancy, and a novel imaging technique called needle-based confocal laser endomicroscopy to observe intraductal structures through a needle, has been reported. An understanding of the morphological characteristics of PCLs depicted by ultrasound imaging and of the benefits and limitations of EUS diagnosis in daily practice is needed.

Glycopatterns and Glycoproteins Changes in MCN and SCN: A Prospective Cohort Study

Background. Advances in imaging improve the detection of malignant pancreatic cystic including mucinous cystic neoplasm (MCN), intraductal papillary mucinous neoplasm (IPMN), and mucinous cystic adenocarcinoma (MCA), but the distinction between benign and malignant lesions remains a problem. In an effort to establish glycopatterns as potential biomarkers for differential diagnosis between MCN and SCN, we systematically investigated the alterations of glycopatterns in cystic fluids for both SCN and MCN. Methods. Among the 75 patients enrolled, 37 were diagnosed as MCN and 38 as SCN based on histology. Lectin microarray analysis was performed on each sample, and the fluorescence intensity was used to obtain the fold-change. Then, mixed cyst fluids of MCN group and SCN group were cross bonded with magnetic particles coupled by Lectin STL and WGA, respectively. Hydrophilic interaction liquid chromatography (HILIC) enrichment was performed, liquid chromatography (LC)/mass spectrometry (MS) analysis and bioinformatical analysis was conducted to find the differential glycoproteins between MCNs and SCNs. Results. Through analysis of lectin microarray between MCNs and SCNs, stronger lectin signal patterns were assigned to Lectin WFA, DBA, STL, WGA, and BPL; and weaker signal patterns were assigned to Lectin PTL-I, Con A, ACA, and MAL-I. The glycoproteins were enriched by STL or WGA-coupled magnetic particles. Furthermore, the 10 identified correspondding genes were found to be significantly elevated in the mucinous cystadenoma: CLU, A2M, FGA, FGB, FGG, PLG, SERPINA1, SERPING1, C5, C8A, and C9. Bioinformatics analysis revealed that the above genes may activate the KEGG pathway: immune complement system. Conclusion. This study shows changes in glycopatterns and glycoproteins are associated with MCNs and SCNs.

A prospective study on endoscopic ultrasound for the differential diagnosis of serous cystic neoplasms and mucinous cystic neoplasms

Background

To provide criteria for the differential diagnosis of serous cystic neoplasms (SCNs) and mucinous cystic neoplasms (MCNs) by analyzing the imaging features of these two neoplasms by endoscopic ultrasound (EUS).

Methods

From April 2015 to December 2017, a total of 69 patients were enrolled in this study. All patients were confirmed to have MCNs (31 patients) or SCNs (38 patients) by surgical pathology. All patients underwent EUS examination. The observation and recorded items were size, location, shape, cystic wall thickness, number of septa, and solid components.

Results

Head/neck location, lobulated shape, thin wall and > 2 septa were the specific imaging features for the diagnosis of SCNs. When any two imaging features were combined, we achieved the highest area under the curve (Az) (0.824), as well as the appropriate sensitivity (84.2%), specificity (80.6%), positive predictive value (PPV) (84.2%), and negative predictive value (NPV) (80.6%). Body/tail location, round shape, thick wall and 0–2 septa were the specific imaging features for the diagnosis of MCNs. When any three imaging features were combined, we obtained the highest Az value (0.808), as well as the appropriate sensitivity (77.4%), specificity (84.2%), PPV (80.0%) and NPV (82.1%).

Conclusions

Pancreatic cystadenomas that meet any two of the four imaging features of head/neck location, lobulated shape, thin wall and > 2 septa could be diagnosed as SCNs, and those that meet any three of the four imaging features of body/tail location, round shape, thick wall and 0–2 septa could be considered as MCNs.

Trial registration

The study was registered at the Chinese Clinical Trial Registry. The registration identification number is ChiCTR-OOC-15006118. The date of registration is 2015-03-20.

An international, multi-institution survey of the use of EUS in the diagnosis of pancreatic cystic lesions

Background and Objectives:

Currently, pancreatic cystic lesions (PCLs) are recognized with increasing frequency and have become a more common finding in clinical practice. EUS is challenging in the diagnosis of PCLs and evidence-based decisions are lacking in its application. This study aimed to develop strong recommendations for the use of EUS in the diagnosis of PCLs, based on the experience of experts in the field.

Methods:

A survey regarding the practice of EUS in the evaluation of PCLs was drafted by the committee member of the International Society of EUS Task Force (ISEUS-TF). It was disseminated to experts of EUS who were also members of the ISEUS-TF. In some cases, percentage agreement with some statements was calculated; in others, the options with the greatest numbers of responses were summarized.

Results:

Fifteen questions were extracted and disseminated among 60 experts for the survey. Fifty-three experts completed the survey within the specified time frame. The average volume of EUS cases at the experts' institutions is 988.5 cases per year.

Conclusion:

Despite the limitations of EUS alone in the morphologic diagnosis of PCLs, the results of the survey indicate that EUS-guided fine-needle aspiration is widely expected to become a more valuable method.