AI Tool Analyzes Capsule Endoscopy Videos for Accurately Predicting Patient Outcomes for Crohn’s Disease

There is a clear need for reliable predictors of disease prognosis and response to treatment. Capsule endoscopy, which uses a tiny device fitted with a camera and transmitter to analyze the entire digestive system, offers a potential solution. However, each endoscopy capsule film yields about 10 – 12 thousand images, making it challenging for doctors to identify all crucial details due to the high volume of visual data. Now, a newly developed artificial intelligence (AI) algorithm can rapidly scan and thoroughly analyze all the images produced by the endoscopic capsule within minutes. It can spot inflammations and ulcers, as well as gauge the severity of Crohn’s Disease to help physicians select the best treatments.

Sheba Medical Center (Ramat Gan, Israel) has partnered with Intel (Santa Clara, CA, USA) to develop the AI algorithm-based application that scans and analyzes the images generated by the endoscopic capsule, detecting signs of inflammation and ulcers and evaluating the severity of the disease to inform treatment decisions. The AI platform is expected to reduce the risk of medical complications and hospital admissions, as well as the necessity for invasive procedures, owing to its ability to facilitate early detection of Crohn's Disease.

A recent peer-reviewed study verified the accuracy of the AI algorithm in predicting treatment outcomes for Crohn's Disease, highlighting its potential in guiding the decision for biological therapy. In the study, a team of data researchers and physicians tested this deep learning model using complete capsule endoscopy videos from 101 patients with Crohn's Disease to assess its predictive accuracy for biological therapy. The AI algorithm demonstrated an impressive 81% accuracy, significantly outperforming the analysis by a gastroenterologist based on the inflammatory index in stool samples (calprotectin).

This latest research comes on the back of a trial last year in which the AI algorithm demonstrated its ability to process up to 12,000 images in roughly two minutes. Moreover, it proved to be an extremely effective diagnostic tool, offering 86% accuracy in image and data analysis, compared to the 68% accuracy achieved by an experienced gastroenterologist. The AI analysis also compared favorably to the analysis based on the inflammatory index in stool (calprotectin).

“Predicting disease course and patient outcomes for Crohn’s Disease is one of the most critical clinical challenges in inflammatory bowel disease treatment. However, this research highlights the potential impact of AI on this process,” said Prof. Uri Kopylov, Director of IBD in the Department of Gastroenterology at Sheba. “By adopting AI in clinical practice, we can begin to use our wealth of knowledge and research in personalized medicine to drive improved patient outcomes and open the door to new possibilities for diagnosis and treatment.”

Related Links:
Sheba Medical Center
Intel