Artificial intelligence-based diagnosis of upper gastrointestinal subepithelial lesions on endoscopic ultrasonography images.

BACKGROUND: Endoscopic ultrasonography (EUS) is useful for the differential diagnosis of subepithelial lesions (SELs); however, not all of them are easy to distinguish. Gastrointestinal stromal tumors (GISTs) are the commonest SELs, are considered potentially malignant, and differentiating them from benign SELs is important. Artificial intelligence (AI) using deep learning has developed remarkably in the medical field. This study aimed to investigate the efficacy of an AI system for classifying SELs on EUS images. METHODS: EUS images of pathologically confirmed upper gastrointestinal SELs (GIST, leiomyoma, schwannoma, neuroendocrine tumor [NET], and ectopic pancreas) were collected from 12 hospitals. These images were divided into development and test datasets in the ratio of 4:1 using random sampling; the development dataset was divided into training and validation datasets. The same test dataset was diagnosed by two experts and two non-experts. RESULTS: A total of 16,110 images were collected from 631 cases for the development and test datasets. The accuracy of the AI system for the five-category classification (GIST, leiomyoma, schwannoma, NET, and ectopic pancreas) was 86.1%, which was significantly higher than that of all endoscopists. The sensitivity, specificity, and accuracy of the AI system for differentiating GISTs from non-GISTs were 98.8%, 67.6%, and 89.3%, respectively. Its sensitivity and accuracy were significantly higher than those of all the endoscopists. CONCLUSION: The AI system, classifying SELs, showed higher diagnostic performance than that of the experts and may assist in improving the diagnosis of SELs in clinical practice.

Usefulness of endoscopic ultrasound-guided fine-needle biopsy for the diagnosis of autoimmune pancreatitis using a 22-gauge Franseen needle: a prospective multicenter study.

BACKGROUND: Detailed histological evaluation is important in the diagnosis of autoimmune pancreatitis (AIP). However, it remains challenging to obtain adequate tissue from the pancreas. Recently, several reports have suggested the usefulness of endoscopic ultrasound-guided fine-needle biopsy (EUS-FNB) using the new "core" needles for acquiring pancreatic tissue. We aimed to investigate the usefulness of EUS-FNB for diagnosing AIP with one such needle, a 22-gauge Franseen needle. METHODS: Patients who met the imaging diagnostic criteria for AIP based on the International Consensus Diagnostic Criteria (ICDC) were enrolled in the study. All patients underwent EUS-FNB with a 22-gauge Franseen needle. Histological findings were evaluated based on the ICDC, and the detection rates of level 1 and level 1 or 2 histology were calculated. RESULTS: 56 patients from 11 different institutions were enrolled in the final analysis (55 suspected to have type 1 AIP and one with type 2 AIP). Lymphoplasmacytic infiltration, obliterative phlebitis, storiform fibrosis, and > 10 IgG4-positive cells per high-power field were detected in 55 (100 %), 24 (43.6 %), 40 (72.7 %), and 36 (65.5 %) of the 55 patients, respectively. The detection rates of level 1 and level 1 or 2 histology for AIP were 58.2 % (95 % confidence interval [CI] 44.1 % - 71.3 %) and 92.7 % (95 %CI 82.4 % - 98.0 %), respectively, which were apparently higher than our historical results (7.9 % [95 %CI 1.7 % - 21.4 %] and 62.2 % [95 %CI 46.5 % - 76.2 %], respectively) using a conventional needle. CONCLUSIONS: EUS-FNB with a 22-gauge Franseen needle demonstrated favorable detection rates which would be clinically beneficial for the histological diagnosis of AIP.

Quantitative evaluation of pancreatic tumor fibrosis using shear wave elastography.

BACKGROUND & AIMS: There is no established non-invasive method for diagnosis of pancreatic fibrosis. Shear wave elastography (SW-EG) may be a candidate for this purpose. The aims of this study were to assess the reproducibility of SW-EG in the normal imaging pancreas (Phase 1) and to evaluate the diagnostic performance of SW-EG for pancreatic fibrosis classified histologically (Phase 2). METHODS: Phase 1: This included 127 cases that underwent SW-EG of the normal imaging pancreas. SW-EG was measured at least five times in the pancreatic parenchyma and the median of repeated measurements was defined as the pancreatic elastic modulus (PEM). Phase 2: This included 53 cases that underwent SW-EG of the pancreatic parenchyma preoperatively and in which pancreas parenchyma were evaluated histologically. Histological fibrosis was graded in 4 stages: normal, mild, moderate, and severe. RESULTS: Phase 1: Median PEM in the head, body, and tail of the pancreas were 3.23, 3.17, and 2.91 kPa, respectively, with no significant difference among regions (P = 0.554). The intraclass correlation coefficient showed good reproducibility (ρ = 0.71) after 5 measurements. Phase 2: There was a significant positive correlation between PEM and the histological pancreatic fibrosis stage (rs = 0.63, P < 0.001). Areas under the receiver operating characteristic curve for the accuracy of SW-EG for diagnosis of pancreatic fibrosis were 0.85 (≥mild), 0.84 (≥moderate), and 0.87 (severe). CONCLUSION: SW-EG can be used to determine the stage of pancreatic fibrosis non-invasively with high accuracy and reproducibility.

The prognostic factors and trajectory of HRQOL in patients with pancreatic cancer who received psychiatric intervention.

BACKGROUND AND AIM: Patients with advanced pancreatic cancer have severe pain, anxiety, and depression, and these symptoms deteriorate quality of life (QOL). Previous study reported that early psychiatric intervention for advanced cancer patients may improve QOL. We evaluated the trajectory of health-related QOL (HRQOL) in patients with pancreatic cancer after an early psychiatric intervention. METHODS: A prospective cohort study was performed in 108 pancreatic cancer patients who received consultation liaison psychiatry from November, 2011 to October, 2014 at Nagoya University Hospital. Longitudinal changes in HRQOL and the association between aspects of HRQOL and survival were evaluated using European Organization for Research and Treatment of Cancer Quality of Life Questionnaire core 30 before treatment and every month after the start of treatment. RESULTS: Physical functioning significantly decreased after 1, 3, and 6 months of treatment compared with baseline, but global health status (GHS), role functioning, emotional functioning, and social functioning showed a tendency to improve. Severe impairment of GHS, fatigue, and appetite loss were apparent at 3 months prior to death. In multivariate analysis, pain was a significant prognostic factor for survival (hazard ratio [HR], 1.109; 95% confidence interval [CI], 1.021-1.204; P = 0.013), in addition to poor performance status (HR, 5.473; 95%CI, 2.338-12.816; P < .0001) and distant metastases at diagnosis (HR, 3.274; 95%CI, 1.872-5.725; P < .0001). CONCLUSIONS: Early psychiatric intervention in patients with pancreatic cancer may maintain HRQOL. More effective pain management and reduction of psychological distress are important for patients with pancreatic cancer.

Functional Correlation Between the Pancreas and the Small Intestine in Humans: The First Evaluation Using a Newly Developed Enteroscopy.

OBJECTIVES: The aim of this study is to evaluate a functional correlation between the pancreas and the small intestine and the association of this relationship with nutritional status, using magnifying enteroscopy. METHODS: The subjects were adults aged 20 years or older who underwent upper gastrointestinal endoscopy. An endoscope was inserted into the jejunum, and 10% glucose was sprayed under magnifying observation to evaluate changes in blood flow in the villous capillary network. Mucosal biopsy was performed before and after spraying to evaluate the incretin response in the jejunal mucosa. RESULTS: A total of 124 patients participated in the study. There was a positive correlation between villous blood flow change and exocrine pancreas function (R = 0.4337, P < 0.0001). Changes of gastric inhibitory polypeptide and glucagon-like peptide messenger RNAs in biopsy samples were positively correlated with endocrine pancreas function in 88 patients without treatment for diabetes (R = 0.4314, P = 0.0012; R = 0.4112, P = 0.0081). In patients with lower villous blood flow change and decreased pancreatic exocrine function, the prognostic nutritional index were significantly lower (P = 0.0098), compared with other patients. CONCLUSIONS: This study provides the first evidence of a close functional correlation between the pancreas and the small intestine.

Age-related changes in pancreatic elasticity: When should we be concerned about their effect on strain elastography?

BACKGROUND: Ultrasound strain elastography is one of the useful methods for evaluating pancreatic lesions. During aging, several pancreatic parenchymal changes occur that may interfere with the interpretation of the ultrasound images. We studied age-related changes in pancreatic elasticity using transabdominal ultrasound strain elastography in subjects without known pancreatic disease. METHODS: This study was conducted at Nagoya University Hospital, which is an academic medical center, and included 102 subjects (66 women and 39 men) aged 20-85years (mean 58.6±17.5) who underwent transabdominal ultrasonography for screening and follow-up for non-pancreatic diseases. Strain elastography of the pancreas was performed, and the results were subjected to quantitative strain histogram analysis. The correlations of age with four elastographic parameters (Mean, Standard deviation, Skewness, and Kurtosis) and other findings, including hyperechoic pancreas, hyperechoic liver, and diabetes, were evaluated. RESULTS: There was a significant correlation between increasing age and elastographic parameters such as the Mean (P=0.004), Skewness (P=0.007), and Kurtosis (P=0.03), and these differences became significant after the age of 40. The prevalence of hyperechoic pancreas increased with age (P<0.001), and the Means were lower in those with hyperechoic pancreas (P=0.004) and a higher body mass index (BMI, P=0.008). No significant correlations with diabetes, hyperechoic liver, or elastographic parameters were demonstrated. CONCLUSION: Strain elastography demonstrated elastographic changes in the pancreas with aging that included a decreasing Mean and increasing Skewness and Kurtosis after the age of 40. The prevalence of pancreatic hyperechogenicity increased, and the pancreatic hyperechogenicity was significantly negatively correlated with the Mean.

Investigation of Morphological and Functional Changes in the Small Intestine With Pancreatic Disease.

OBJECTIVES: The aim of this study was to investigate the relationship between pancreas and small intestine evaluating the endoscopic and histopathologic findings of the proximal small intestine in pancreatic diseases. METHODS: Fifty patients (18 patients with chronic pancreatitis, 17 patients with pancreatic cancer, 15 control subjects) underwent enteroscopy using a prototype enteroscope. The villous height of the jejunum on bioptic specimens was measured, and the mean values of the villi were compared among the 3 groups. Exocrine function was calculated by the pancreatic function diagnostic test, and the correlation between the recovery rate of p-aminobenzoic acid and the villous height was assessed. Finally, the distribution of the K cells secreting glucose-dependent insulinotropic polypeptide and the L cells secreting glucagon-like peptide 1 in the duodenum and jejunum was investigated using immunohistochemistry for glucose-dependent insulinotropic polypeptide and glucagon-like peptide 1. RESULTS: The mean villous height in chronic pancreatitis (328 ± 67 µm) was significantly lower than that in pancreatic cancer (413 ± 57 µm) and control subjects (461 ± 97 µm) (P = 0.004 and P < 0.0001, respectively). A positive correlation was found between the recovery rate of p-aminobenzoic acid and the villous height (r = 0.52, P = 0.0001). The presence of K and L cells was verified in the duodenum and the jejunum. CONCLUSIONS: Close relationship between pancreas and small intestine was demonstrated.

Quantitative analysis of diagnosing pancreatic fibrosis using EUS-elastography (comparison with surgical specimens).

BACKGROUND: An accurate diagnosis of pancreatic fibrosis is clinically important and may have potential for staging chronic pancreatitis. The aim of this study was to diagnose the grade of pancreatic fibrosis through a quantitative analysis of endoscopic ultrasound elastography (EUS-EG). METHODS: From September 2004 to October 2010, 58 consecutive patients examined by EUS-EG for both pancreatic tumors and their upstream pancreas before pancreatectomy were enrolled. Preoperative EUS-EG images in the upstream pancreas were statistically quantified, and the results were retrospectively compared with postoperative histological fibrosis in the same area. For the quantification of EUS-EG images, 4 parameters (mean, standard deviation, skewness, and kurtosis) were calculated using novel software. Histological fibrosis was graded into 4 categories (normal, mild fibrosis, marked fibrosis, and severe fibrosis) according to a previously reported scoring system. RESULTS: The fibrosis grade in the upstream pancreas was normal in 24 patients, mild fibrosis in 19, marked fibrosis in 6, and severe fibrosis in 9. Fibrosis grade was significantly correlated with all 4 quantification parameters (mean r = -0.75, standard deviation r = -0.54, skewness r = 0.69, kurtosis r = 0.67). According to the receiver operating characteristic analysis, the mean was the most useful parameter for diagnosing pancreatic fibrosis. Using the mean, the area under the ROC curves for the diagnosis of mild or higher-grade fibrosis, marked or higher-grade fibrosis and severe fibrosis were 0.90, 0.90, and 0.90, respectively. CONCLUSIONS: An accurate diagnosis of pancreatic fibrosis may be possible by analyzing EUS-EG images.

Preliminary study on evaluation of the pancreatic tail observable limit of transabdominal ultrasonography using a position sensor and CT-fusion image.

BACKGROUND AND AIM: Transabdominal ultrasonography (US) is commonly used for the initial screening of bilio-pancreatic diseases in Asian countries due to its widespread availability, the non-invasiveness and the cost-effectiveness. However, it is considered that US has limits to observe the area, namely the blind area. The observation of the pancreatic tail is particularly difficult. The goal of this study was to examine the pancreatic tail region that cannot be visualized on transverse scanning of the upper abdomen using US with spatial positional information and factors related to visualization, and observation of the tail from the splenic hilum. METHODS: Thirty-nine patients with pancreatic/biliary tract disease underwent CT and US with GPS-like technology and fusion imaging for measurement of the real pancreatic length and the predicted/real unobservable (PU and RU) length of the pancreatic tail. RU from US on transverse scanning and the real pancreatic length were used to determine the unobservable area (UA: RU/the real pancreatic length). Relationships of RU with physical and hematological variables that might influence visualization of the pancreatic tail were investigated. RESULTS: The real pancreatic length was 160.9 ± 16.4mm, RU was 41.0 ± 17.8mm, and UA was 25.3 ± 10.4%. RU was correlated with BMI (R=0.446, P=0.004) and waist circumferences (R=0.354, P=0.027), and strongly correlated with PU (R=0.788, P<0.001). The pancreatic tail was visible from the splenic hilum in 22 (56%) subjects and was completely identified in 13 (33%) subjects. CONCLUSIONS: Combined GPS-like technology with fusion imaging was useful for the objective estimation of the pancreatic blind area.