Centro Hospitalar de São João Porto, Porto, Portugal
Tiago Ribeiro, MD1, Miguel Mascarenhas, MD1, Joao Afonso, MD1, João Ferreira, PhD2, Hélder Cardoso, MD1, Patrícia Andrade, MD1, Marco Parente, PhD2, Renato Natal, PhD2, Guilherme Macedo, MD, PhD, FACG1 1Centro Hospitalar de São João, Porto, Porto, Portugal; 2University of Porto, Porto, Porto, Portugal
Introduction: Small bowel ulcers and erosions are frequent findings and can have multiple etiologies, including Crohn’s disease, non-steroid anti-inflammatory drugs enteropathy, infections, or small bowel neoplasia. Double-balloon enteroscopy (DBE) allows deep exploration of the small bowel, enabling tissue sampling and the application of endoscopic therapy. The application of artificial intelligence (AI) algorithms to different endoscopic techniques has provided promising results. Convolutional Neural Networks (CNN) are a highly efficient multi-layered artificial intelligence architecture designed for image analysis. The application of these tools for automatic detection of lesions in DBE has not been explored. We aimed to develop and test a CNN-based algorithm for automatic detection of ulcers and erosions in DBE exams.
Methods: A convolutional neural network was developed based on 72 DBE exams. A total of 5805 images were included, 450 images containing ulcers or erosions. The remaining images showed normal mucosa (n=5355). A training dataset comprising 80% of the total pool of images (n=4644) was used for development of the network. The performance of the CNN was evaluated using an independent validation dataset (20% of total image pool, n=1161). The output provided by the network was compared to a consensus classification provided by two gastroenterologists with experience in DBE (Figure 1a). The sensitivity, specificity, accuracy, positive and negative predictive values (PPV and NPV, respectively), and area under the curve (AUC) were calculated.
Results: Our model automatically detected small bowel ulcers or erosions with an accuracy of 96.6% (Figure 1b). The sensitivity, specificity, PPV and NPV were, respectively, 100.0%, 96.4%, 69.8%, and 100.0%. The AUC was 1.00 (Figure 1c). The CNN analyzed the validation dataset in 5 seconds, at a rate of approximately 241 frames per second.
Discussion: The authors developed, for the first time, an algorithm for automatic detection of enteric ulcers or erosions during DBE exams. The development and application of these computational tools to advanced enteroscopy techniques may further enhance their diagnostic yield, which may contribute to more efficient diagnostic processes, ultimately improving the management of these patients.
Figure: Figure 1: 1a - Output obtained from the application of the convolutional neural network. A blue bar represents a correct prediction. 1b - Evolution of the accuracy of the convolutional neural network during training and validation phases, as the training and validation datasets were repeatedly inputted in the neural network. 1c - ROC analyses of the network’s performance. N: normal mucosa; UE – ulcers or erosions.
Disclosures: Tiago Ribeiro indicated no relevant financial relationships. Miguel Mascarenhas indicated no relevant financial relationships. Joao Afonso indicated no relevant financial relationships. João Ferreira indicated no relevant financial relationships. Hélder Cardoso indicated no relevant financial relationships. Patrícia Andrade indicated no relevant financial relationships. Marco Parente indicated no relevant financial relationships. Renato Natal indicated no relevant financial relationships. Guilherme Macedo indicated no relevant financial relationships.
Tiago Ribeiro, MD1, Miguel Mascarenhas, MD1, Joao Afonso, MD1, João Ferreira, PhD2, Hélder Cardoso, MD1, Patrícia Andrade, MD1, Marco Parente, PhD2, Renato Natal, PhD2, Guilherme Macedo, MD, PhD, FACG1. P1965 - Automatic Detection of Small Bowel Ulcers and Erosions in Double-Balloon Enteroscopy Using a Convolutional Neural Network: A Proof-of-Concept Study, ACG 2021 Annual Scientific Meeting Abstracts. Las Vegas, Nevada: American College of Gastroenterology.