ME-APDS; MAGENTIQ-COLO

The Magentiq Eye Automatic Polyp Detection System (ME-APDS™) is an adjunctive tool leveraging deep learning algorithms to analyze real-time colonoscopy video, aiming to enhance the identification of gastrointestinal lesions. Its methodology involves processing video frames to detect and highlight suspicious mucosal abnormalities on a display, thereby assisting endoscopists in improving polyp detection rates, particularly for adenomas.

The Magentiq Eye Automatic Polyp Detection System (ME-APDS™), also known as MAGENTIQ-COLO™, is an adjunctive software-based medical device designed to enhance video colonoscopy procedures for endoscopists during screening and surveillance. Its core function is to assist in the real-time identification of mucosal abnormalities, primarily polyps, by analyzing the digital video stream from a standard white-light endoscopy camera; the system leverages deep machine learning algorithms, specifically Deep Neural Networks (DNN), alongside other supporting computational methods to process video frames, and upon detecting regions exhibiting visual characteristics consistent with polyps, the ME-APDS highlights these areas as an overlay on the original video, displayed on a dedicated touchscreen or other monitors in the procedure room, while being explicitly intended as an aid and not as a standalone diagnostic tool, nor does it replace the need for histopathological sampling for definitive diagnosis. The device's performance was evaluated through comprehensive testing; non-clinical assessments verified its image quality, software robustness, electromagnetic compatibility, and electrical safety, noting a median marker annotation latency of 0.166 seconds for polyps, and standalone algorithm testing on 172 colonoscopy videos (with 449 polyps) showed high polyp-wise recall (90%-100% depending on criteria) and a median 82% coverage for histologically verified polyps, meeting predefined false positive rates, while a pivotal multi-center, randomized controlled clinical trial with 950 patients demonstrated that ME-APDS-assisted colonoscopy (MEAC) significantly increased the Adenomas Per Colonoscopy (APC) by 37% (0.70 for MEAC vs. 0.51 for conventional colonoscopy) and was non-inferior in Adenomas Per Extraction (APE) (0.31 vs. 0.27), with the MEAC arm showing superior detection of smaller (≤5mm), larger (>6-9mm), sessile, and flat polyps, particularly in the proximal colon, contributing to a higher overall adenoma detection rate (ADR) and a four-month earlier recommendation for follow-up colonoscopies, without increasing adverse events or procedure duration.

AI Regulatory Concerns

Powered by

Jev reads the full FDA letter and returns calibrated concern probabilities per regulatory category. These are AI estimates, not regulatory advice.

• Q1. How does the ME-APDS algorithm differentiate between true polyps and polyp-like mimics (e.g., folds, debris, poor bowel preparation) to maintain a low false positive rate under diverse clinical conditions?; What were the specific performance metrics (sensitivity, specificity, positive predictive value) for detecting different polyp morphologies (e.g., sessile, pedunculated, flat) and histologies (e.g., adenomatous vs. hyperplastic) in the clinical trial?; The standalone testing involved 172 videos. How were these videos selected, and do they represent the full spectrum of image quality, bowel preparation, and procedural variations encountered in routine clinical practice?; Beyond increasing Adenoma Detection Rate (ADR), what is the system's impact on detecting clinically significant lesions, such as advanced adenomas or sessile serrated lesions, particularly those easily missed by the human eye?; The document mentions a 4-month earlier next scheduled colonoscopy for the MEAC cohort. What were the specific criteria or risk stratification changes driven by MEAC findings that led to this shortened surveillance interval recommendation?