Colon capsule endoscopy · Panendoscopic AI
One capsule. The full GI tract. Every finding in view.
Panendoscopic AI assistance for detecting blood and mucosal lesions throughout colon capsule endoscopy examinations.
COLON CAPSULE · DUAL CAMERA GI SPY ACTIVE
CAMERA APOLYP · 94%
CAMERA BNORMAL MUCOSA · 98%
ESOPHAGUSSTOMACHSMALL BOWELCOLON
NORMAL MUCOSAREVIEWEDPOLYPFLAGGEDBLOODREVIEWEDVASCULAR LESIONREVIEWED
ANATOMY: COLONCLASSIFICATION: POLYP
01 · Intended use
Panendoscopic AI assistance for colon capsule review.
GI Spy® supports post-procedural review of colon capsule endoscopy examinations spanning the gastrointestinal tract.
It detects blood and mucosal lesions and brings relevant frames into focus for physician validation.
AI
The clinical challenge
One examination can contain up to 50,000 frames.
Colon capsule endoscopy provides a non-invasive view across the gastrointestinal tract, but its scale makes manual review lengthy and relevant findings easier to overlook.
“Panendoscopic value depends on turning a vast examination into a clear, clinically navigable review.”
02
How it works
One continuous path from capsule video to validated finding.
GI Spy® supports the clinician across the complete panendoscopic review while keeping final interpretation in expert hands.
Exam acquisition
A previously acquired capsule endoscopy video enters the review workflow.
Panendoscopic analysis
The AI analyzes frames captured across the esophagus, stomach, small bowel and colon.
Finding detection
Blood and clinically relevant mucosal lesions are identified and categorized.
Prioritized review
Relevant findings are assembled into a focused, navigable review sequence.
Clinical validation
The clinician confirms findings in anatomical and clinical context.
AI capabilities
Designed for broad colon capsule lesion review.
97.6%
Research frame-level accuracy
Panendoscopic detection
Supports lesion review throughout the gastrointestinal tract in colon capsule examinations.
Blood and mucosal lesions
Research validation differentiated normal mucosa, luminal blood and multiple colonic mucosal lesion types.
Broad lesion coverage
The validated image set included protruding, vascular, ulcerative and erosive findings.
High-speed frame analysis
The research CNN processed approximately 65 frames per second in the validation dataset.
Focused review
Relevant frames can be surfaced from examinations containing as many as 50,000 images.
03
Scientific validation
Research built specifically around colon capsule imaging.
Deep learning and colon capsule endoscopy: automatic detection of blood and colonic mucosal lesions using a convolutional neural network
Ribeiro T, et al. · 2022 · Frame-level development and validation study
Read publication ↗Clinical benefits
The complete GI tract. A focused review.
01
Faster analysis of long colon capsule examinations
↗02
Detection support across the full gastrointestinal tract
↗03
Reduced risk of overlooking blood or clinically relevant mucosal lesions
↗04
A non-invasive pathway with potential to support colorectal screening workflows
↗GI Spy®
See the whole tract. Focus on what matters.
Talk to our team ↗GI Spy® is part of DigestAID’s research and development pipeline. Availability and regulatory status may vary by market.