Artificial intelligence using a convolutional neural network for automatic detection of small‐bowel angioectasia in capsule endoscopy images

Abstract

Background and Aim Although small‐bowel angioectasia is reported as the most common cause of bleeding in patients and frequently diagnosed by capsule endoscopy (CE) in patients with obscure gastrointestinal bleeding, a computer‐aided detection method has not been established. We developed an artificial intelligence system with deep learning that can automatically detect small‐bowel angioectasia in CE images. Methods We trained a deep convolutional neural network (CNN) system based on Single Shot Multibox Detector using 2237 CE images of angioectasia. We assessed its diagnostic accuracy by calculating the area under the receiver operating characteristic curve (ROC‐AUC), sensitivity, specificity, positive predictive value, and negative predictive value using an independent test set of 10 488 small‐bowel images, including 488 images of small‐bowel angioectasia. Results The AUC to detect angioectasia was 0.998. Sensitivity, specificity, positive predictive value, and negative predictive value of CNN were 98.8%, 98.4%, 75.4%, and 99.9%, respectively, at a cut‐off value of 0.36 for the probability score. Conclusions We developed and validated a new system based on CNN to automatically detect angioectasia in CE images. This may be well applicable to daily clinical practice to reduce the burden of physicians as well as to reduce oversight.

A pilot trial of Convolution Neural Network for automatic retention-monitoring of capsule endoscopes in the stomach and duodenal bulb

“…If the CE passes through the stomach and duodenal bulb into the DSD too slowly, the medical staff have to be continuously monitoring even for hours, and they have to discern whether the long retention of the CE occurred in the stomach for fear of the battery exhausted. Now a computational method has been developed for detection of a broader spectrum of the lesions in the small bowels [19][20][21][22][23] , such as obscure gastrointestinal bleeding, erosions, ulcers, and angioectasia. From the results of this research, the sensitivity and specificity was >85% on the whole.…”

“…How to predict the residence time of the CE in the stomach or duodenal bulb has not been solved, and medical staff may have to wait for several hours in the examination room to monitor whether the CE enters the descending segment of the duodenum (DSD) 5,6 . If the CE cannot enter the DSD in 2-3 h, some interventions, e.g., drugs or gastroscopy, can be used to push the CE forward into the DSD 7 , which is a tedious and boring task, especially for some patients who have to undergo the CE examination at the same time, which could greatly increase the monitoring workload for the medical staff.Artificial intelligence (AI), as a new technique, has been developed in the recent years, which includes Autoencoder 8 , Deep Belief Network 9 , Convolution Neural Network (CNN) 10 , and Deep Residual Network 11 , and they have been used in the medical image analysis and have been proved to be effective in some medical diagnostic fields, such as pulmonary nodules 12 , breast lesions 13,14 , skin cancer 15 , early gastrointestinal cancers 16,17 , polyps 18 , and small-bowel diseases [19][20][21][22][23] .Of those techniques, the CNN 24 is a type of deep learning mode 25-27 that requires the preprocessing of the image data inputted as a training image set for extracting specific features and quantities by using the multiple network layers (convolutional layers, fully-connected layers), and then iteratively changed through the multiple convolutions and the non-linear operations until the training data set is converted into a probability distribution of the potential image categories. With its high efficiency in the image analysis, the CNN has become a principal method of deep learning for images.…”

“…Artificial intelligence (AI), as a new technique, has been developed in the recent years, which includes Autoencoder 8 , Deep Belief Network 9 , Convolution Neural Network (CNN) 10 , and Deep Residual Network 11 , and they have been used in the medical image analysis and have been proved to be effective in some medical diagnostic fields, such as pulmonary nodules 12 , breast lesions 13,14 , skin cancer 15 , early gastrointestinal cancers 16,17 , polyps 18 , and small-bowel diseases [19][20][21][22][23] .…”

A New Active Locomotion Capsule Endoscopy under Magnetic Control and Automated Reading Program

“…When CNN-based interpretations were used for the detection of angioectasia, the most common SB vascular lesion, it demonstrated excellent sensitivity and specificity close to 100%. 45,46 In addition, CNN-based inter-pretations showed an accuracy of 90.8% in the diagnosis of SB erosions and ulcers 47 and showed a higher diagnostic accuracy in blood contents (sensitivity, 96.63%; specificity, 99.96%) compared to conventional suspected blood indicators. 48 In a large-scale study in China, a new CNN interpretation model based on 113,426,569 images was developed, which showed a higher sensitivity (77.9%-99.9%) and lesion detection rate (54.6%-70.9%) compared to conventional CE interpretations for detecting various SB lesions, such as inflammation, ulcers, and polyps.…”

Position statement on priorities for artificial intelligence in GI endoscopy: a report by the ASGE Task Force

“…[2][3][4] Computer vision software is also being evaluated in GI endoscopy to automatically assess bowel preparation quality, to identify possible areas of bleeding during capsule endoscopy, and to facilitate detection of esophageal squamous cancers and dysplastic Barrett's esophagus, among innumerable examples. [5][6][7][8][9] Beyond endoscopy, computer vision is also being applied quite extensively in abdominal radiology, for instance, to classify and risk stratify pancreatic cysts identified on CT and magnetic resonance imaging, in some cases with algorithms that can integrate other clinical and laboratory data as well. 10,11 To focus only on computer vision, however, creates a too-narrow view of the potential for AI in gastroenterology.…”