Search

The increasing incidence of pancreatic cancer will make it the second deadliest cancer in 2030. Imaging based early diagnosis and image guided treatment are emerging potential solutions. Artificial intelligence (AI) can help provide and improve widespread diagnostic expertise and accurate interventional image interpretation. Accurate segmentation of the pancreas is essential to create annotated data sets to train AI, and for computer assisted interventional guidance. Automated deep learning segmentation performance in pancreas computed tomography (CT) imaging is low due to poor grey value contrast and complex anatomy. A good solution seemed a recent interactive deep learning segmentation framework for brain CT that helped strongly improve initial automated segmentation with minimal user input. This method yielded no satisfactory results for pancreas CT, possibly due to a sub-optimal neural network architecture. We hypothesize that a state-of-the-art U-net neural network architecture is better because it can produce a better initial segmentation and is likely to be extended to work in a similar interactive approach. We implemented the existing interactive method, iFCN, and developed an interactive version of U-net method we call iUnet. The iUnet is fully trained to produce the best possible initial segmentation. In interactive mode it is additionally trained on a partial set of layers on user generated scribbles. We compare initial segmentation performance of iFCN and iUnet on a 100CT dataset using dice similarity coefficient analysis. Secondly, we assessed the performance gain in interactive use with three observers on segmentation quality and time. Average automated baseline performance was 78% (iUnet) versus 72% (FCN). Manual and semi-automatic segmentation performance was: 87% in 15 min. for manual, and 86% in 8 min. for iUNet. We conclude that iUnet provides a better baseline than iFCN and can reach expert manual performance significantly faster than manual segmentation in case of pancreas CT. Our novel iUnet architecture is modality and organ agnostic and can be a potential novel solution for semi-automatic medical imaging segmentation in general.

Objectives: Severe acute pancreatitis is associated with peripancreatic morphologic changes as seen on imaging. Uniform communication regarding these morphologic findings is crucial for accurate diagnosis and treatment. For the original 1992 Atlanta classification, interobserver agreement is poor. We hypothesized that for the revised Atlanta classification, interobserver agreement will be better., Methods: An international, interobserver agreement study was performed among expert and nonexpert radiologists (n = 14), surgeons (n = 15), and gastroenterologists (n = 8). Representative computed tomographies of all stages of acute pancreatitis were selected from 55 patients and were assessed according to the revised Atlanta classification. The interobserver agreement was calculated among all reviewers and subgroups, that is, expert and nonexpert reviewers; interobserver agreement was defined as poor (≤0.20), fair (0.21-0.40), moderate (0.41-0.60), good (0.61-0.80), or very good (0.81-1.00)., Results: Interobserver agreement among all reviewers was good (0.75 [standard deviation, 0.21]) for describing the type of acute pancreatitis and good (0.62 [standard deviation, 0.19]) for the type of peripancreatic collection. Expert radiologists showed the best and nonexpert clinicians the lowest interobserver agreement., Conclusions: Interobserver agreement was good for the revised Atlanta classification, supporting the importance for widespread adaption of this revised classification for clinical and research communications.

Objectives: Hereditary pancreatitis (HP) is characterized by recurrent episodes of inflammation of the pancreas. Radiological imaging is used to diagnose HP and to monitor complications. The aim of this study was to describe specific imaging findings in HP., Methods: We retrospectively collected data of HP patients with serial imaging and reviewed all radiological imaging studies (transabdominal ultrasonography, computed tomography, and magnetic resonance imaging)., Results: We included 15 HP patients, with a mean age of 32.5 years (range, 9-61 years) and mean disease duration of 24.1 years (range, 6-42 years). In total, 152 imaging studies were reviewed. Seventy-three percent of patients had a dilated main pancreatic duct (MPD) (width 3.5-18 mm). The MPD varied in size during disease course, with temporary reduction in diameter after drainage procedures. A severe dilated MPD (>10 mm) often coincided with presence of intraductal calcifications (size, 1-12 mm). In 73% of patients, pancreatic parenchyma atrophy occurred, which did not correlate with presence of exocrine or endocrine insufficiency., Conclusions: In HP, the MPD diameter increases with time, mostly without dilated side branches, and is often accompanied by large intraductal calcifications. The size of the MPD is independent of disease state. Atrophy of pancreatic parenchyma is not correlated with exocrine or endocrine insufficiency.

Objective: Endoscopic ultrasonography (EUS) and MRI are promising tests to detect precursors and early-stage pancreatic ductal adenocarcinoma (PDAC) in high-risk individuals (HRIs). It is unclear which screening technique is to be preferred. We aimed to compare the efficacy of EUS and MRI in their ability to detect clinically relevant lesions in HRI., Design: Multicentre prospective study. The results of 139 asymptomatic HRI (>10-fold increased risk) undergoing first-time screening by EUS and MRI are described. Clinically relevant lesions were defined as solid lesions, main duct intraductal papillary mucinous neoplasms and cysts ≥10 mm. Results were compared in a blinded, independent fashion., Results: Two solid lesions (mean size 9 mm) and nine cysts ≥10 mm (mean size 17 mm) were detected in nine HRI (6%). Both solid lesions were detected by EUS only and proved to be a stage I PDAC and a multifocal pancreatic intraepithelial neoplasia 2. Of the nine cysts ≥10 mm, six were detected by both imaging techniques and three were detected by MRI only. The agreement between EUS and MRI for the detection of clinically relevant lesions was 55%. Of these clinically relevant lesions detected by both techniques, there was a good agreement for location and size., Conclusions: EUS and/or MRI detected clinically relevant pancreatic lesions in 6% of HRI. Both imaging techniques were complementary rather than interchangeable: contrary to EUS, MRI was found to be very sensitive for the detection of cystic lesions of any size; MRI, however, might have some important limitations with regard to the timely detection of solid lesions., (Published by the BMJ Publishing Group Limited. For permission to use (where not already granted under a licence) please go to http://www.bmj.com/company/products-services/rights-and-licensing/)

Background & Aims: The prevalence of pancreatic cysts is not known, but asymptomatic pancreatic cysts are diagnosed with increasing frequency. We investigated the prevalence of pancreatic cysts in individuals who were screened by magnetic resonance imaging (MRI) as part of a preventive medical examination., Methods: Data from consecutive persons who underwent abdominal MRI (n = 2803; 1821 men; mean age, 51.1 +/- 10.8 y) at an institute of preventive medical care were included from a prospective database. All individuals had completed an application form including questions about possible abdominal complaints and prior surgery. MRI reports were reviewed for the presence of pancreatic cysts. Original image sets of all positive MRI reports and a representative sample of the negative series were re-assessed by a blinded, independent radiologist., Results: Pancreatic cysts were reported in 66 persons (2.4%; 95% confidence interval, 1.9-3.0); prevalence correlated with increasing age (P < .001). There was no difference in prevalence between sexes (P = .769). There was no correlation between abdominal complaints and the presence of pancreatic cysts (P = .542). Four cysts (6%) were larger than 2 cm and 3 (5%) were larger than 3 cm. Review of the original image sets by the independent radiologist did not significantly change these findings., Conclusions: The prevalence of pancreatic cysts in a large consecutive series of individuals who underwent an MRI at a preventive medical examination was 2.4%. Prevalence increased with age, but did not differ between sexes. Only a minority of cysts were larger than 2 cm., (Copyright 2010 AGA Institute. Published by Elsevier Inc. All rights reserved.)

Background/aims: The current terminology for describing peripancreatic collections in acute pancreatitis (AP) derived from the Atlanta Symposium (e.g. pseudocyst, pancreatic abscess) has shown a very poor interobserver agreement, creating the potential for patient mismanagement. A study was undertaken to determine the interobserver agreement for a new set of morphologic terms to describe peripancreatic collections in AP., Methods: An international, interobserver agreement study was performed: 7 gastrointestinal surgeons, 2 gastroenterologists and 8 radiologists in 3 US and 5 European tertiary referral hospitals independently evaluated 55 computed tomography (CT) scans of patients with predicted severe AP. The percentage agreement [median, interquartile range (IQR)] for 9 clinically relevant morphologic terms was calculated among all reviewers, and separately among radiologists and clinicians. The percentage agreement was defined as poor (<0.50), moderate (0.51-0.70), good (0.71-0.90), and excellent (0.91-1.00)., Results: Overall agreement was good to excellent for the terms collection (percentage agreement = 1; IQR 0.68-1), relation with pancreas (1; 0.68-1), content (0.88; 0.87-1), shape (1; 0.78-1), mass effect (0.78; 0.62-1), loculated gas bubbles (1; 1-1), and air-fluid levels (1; 1-1). Overall agreement was moderate for extent of pancreatic nonenhancement (0.60; 0.46-0.88) and encapsulation (0.56; 0.48-0.69). The percentage agreement was greater among radiologists than clinicians for extent of pancreatic nonenhancement (0.75 vs. 0.57, p = 0.008), encapsulation (0.67 vs. 0.46, p = 0.001), and content (1 vs. 0.78, p = 0.008)., Conclusion: Interobserver agreement for the new set of morphologic terms to describe peripancreatic collections in AP is good to excellent. Therefore, we recommend that current clinically based definitions for CT findings in AP (e.g. pancreatic abscess) should no longer be used., (Copyright 2008 S. Karger AG, Basel and IAP.)