ABSTRACT
A 68-year-old man was admitted with hematochezia. Emergency computed tomography showed multiple diverticula throughout the colon. Initial colonoscopy on day 2 showed no active bleeding, but massive hematochezia on day 3 led to the performance of an emergency endoscopy. Substantial bleeding in the ileocecal area obscured the visual field, making it challenging to view the area around the bleeding site. Two endoscopic band ligations (EBLs) were applied at the suspected bleeding sites. Hemostasis was achieved without active bleeding after EBL. However, the patient developed lower right abdominal pain and fever (39.4°C) on day 6. Urgent computed tomography revealed appendiceal inflammation, necessitating emergency open ileocecal resection for acute appendicitis. Pathological examination confirmed acute phlegmonous appendicitis, with EBLs noted at the appendiceal orifice and on the anal side. This case illustrates the efficacy of EBL in managing colonic diverticular bleeding. However, it also highlights the risk of appendicitis due to EBL in cases of ileocecal hemorrhage exacerbated by poor visibility due to substantial bleeding. Endoscopists need to consider this rare but important complication when performing EBL in similar situations.
ABSTRACT
Objectives: Localized autoimmune pancreatitis is difficult to differentiate from pancreatic ductal adenocarcinoma on endoscopic ultrasound images. In recent years, deep learning methods have improved the diagnosis of diseases. Hence, we developed a special cross-validation framework to search for effective methodologies of deep learning in distinguishing autoimmune pancreatitis from pancreatic ductal adenocarcinoma on endoscopic ultrasound images. Methods: Data from 24 patients diagnosed with localized autoimmune pancreatitis (8751 images) and 61 patients diagnosed with pancreatic ductal adenocarcinoma (20,584 images) were collected from 2016 to 2022. We applied transfer learning to a convolutional neural network called ResNet152, together with our innovative imaging method contributing to data augmentation and temporal data process. We divided patients into five groups according to different factors for 5-fold cross-validation, where the ordered and balanced datasets were created for the performance evaluations. Results: ResNet152 surpassed the endoscopists in all evaluation metrics with almost all datasets. Interestingly, when the dataset is balanced according to the factor of the endoscopists' diagnostic accuracy, the area under the receiver operating characteristic curve and accuracy were highest at 0.85 and 0.80, respectively. Conclusions: It is deduced that image features useful for ResNet152 correlate with those used by endoscopists for their diagnoses. This finding may contribute to sample-efficient dataset preparation to train convolutional neural networks for endoscopic ultrasonography-imaging diagnosis.