Automatic detection of colorectal neoplasias in wireless colon capsule endoscopic images using a deep convolutional neural network.

BACKGROUND AND AIMS Although colorectal neoplasias are the most common abnormalities found in colon capsule endoscopy (CCE), no computer-aided detection method is yet available. We developed an artificial intelligence (AI) system that uses deep learning to automatically detect such lesions in CCE images. METHODS We trained a deep convolutional neural network system based on a Single Shot Multibox Detector using 15,933 CCE images of colorectal neoplasias such as polyps and cancers. We assessed performance by calculating areas under the receiver operating characteristic curves and sensitivities, specificities, and accuracies using an independent test set of 4,784 images including 1,850 images of colorectal neoplasias and 2,934 normal colon images. RESULTS The area under the curve for detection by the AI model of colorectal neoplasias was 0.902. The sensitivity, specificity, and accuracy were 79.0%, 87.0%, and 83.9%, respectively, at a probability cutoff of 0.348. CONCLUSIONS We developed and validated a new AI-based system that automatically detects colorectal neoplasias in CCE images.