AI in Endoscopy: Olympus ENDO-AID and Pentax CADEYE 2026
Technology 2026 · Endoscopy Silo · Trend
Artificial intelligence in endoscopy: ENDO-AID, CADEYE and the future of endoscopic diagnosis
AI-powered CADe and CADx systems detect polyps in real time during colonoscopy, alert to suspicious lesions during gastroscopy and increase the adenoma detection rate — the biggest revolution in endoscopy since the move to HD
🤖
GSMED Imaging technical team — 2026 endoscopy technology
Based on: ESGE Position Statement on Artificial Intelligence in Endoscopy 2023, Hassan 2024 meta-analysis (ENDO-AID vs non-AI), van der Zander 2024 (CADEYE), Repici A et al. 2024 and Colombia endoscopy AI market data.
Artificial intelligence has arrived in digestive endoscopy to stay — and not as a tool of the future, but as a reality available in 2026 in Colombia. Computer-aided detection (CADe) systems analyze the endoscope video in real time, detect shapes, textures and colors consistent with polyps and alert the endoscopist with a green box visible on the monitor. The result, demonstrated in multiple clinical trials: more adenomas detected, especially flat and difficult ones.
1. CADe vs CADx — two different types of AI in endoscopy
🟩
CADe — Computer Aided Detection
Detects and highlights suspicious lesions in real time with a colored box or square on the monitor — says "there is something here". It does not characterize the lesion or predict if it is benign or malignant. It only detects. Olympus ENDO-AID and Pentax CADEYE are CADe systems. More robust evidence — randomized trials show a real increase in ADR.
→ Detects · Highlights · No characterization · ADR +12–25%
🔎
CADx — Computer Aided Diagnosis
Characterizes the detected lesion — predicts if a polyp is an adenoma or hyperplastic, if a gastric lesion is low or high risk, or if a Barrett area has dysplasia. It goes beyond "there is something" to "this is what". The evidence is more nascent than in CADe and its adoption is slower, but the most advanced systems in 2026 already offer both functions integrated.
→ Characterizes · Predicts histology · Evidence in development
2. ENDO-AID (Olympus) vs CADEYE (Pentax) — comparison
| Feature | ENDO-AID (Olympus) | CADEYE (Pentax) |
|---|---|---|
| Manufacturer | Olympus Medical | Pentax Medical / HOYA |
| Available with | EVIS X1 (CV-1500) · EVIS EXERA III (CV-190) | EPKi+ with i-scan OE |
| Main function | CADe for colonoscopy (polyp detection) | CADe for colonoscopy (adenoma detection) |
| ADR evidence | Hassan 2024: ADR +10–15% vs non-AI (pooled) | van der Zander 2024: ADR +12–15% (most recent meta-analysis) |
| Real-time detection | Yes · <40 ms latency | Yes · <40 ms latency |
| CE/FDA regulation | CE marking · FDA clearance | CE marking · FDA clearance |
| Hardware integration | Only on Olympus equipment — does not work with other brands | Only on Pentax equipment — does not work with other brands |
| Available in Colombia | Yes — GSMED distribution | Yes — GSMED distribution |
3. What AI cannot replace yet
The correct role of AI in endoscopy — assistant, not substitute
Current CADe systems reduce missed lesions (miss rate), especially flat and small adenomas that less experienced endoscopists tend to overlook. But they have clear limitations in 2026: (1) they cannot replace the quality of bowel preparation — AI does not see what fecal residue covers; (2) they have frequent false positives (they flag bile residue, haustral shadows or bubbles as "lesions") — the endoscopist must discern in seconds if the alert is valid; (3) they do not have a sufficiently validated CADx function to decide if a detected polyp needs a biopsy or can be left untreated. The most useful AI in 2026 is the one that has an expert endoscopist who knows when to listen to it and when to ignore the alert.
Does AI in endoscopy increase the procedure time?
Not significantly. ENDO-AID and CADEYE CADe systems have an image processing latency of less than 40 milliseconds — the endoscopist does not even notice. The time increase comes from clinical logic: CADe detects more suspicious lesions, the endoscopist reviews them more carefully and eventually performs more biopsies or polypectomies than without AI. That additional time is clinically desirable — it is the time invested in not overlooking an adenoma. The mean withdrawal time in colonoscopy with CADe vs without CADe is comparable, according to the randomized trials of 2023–2024.
ENDO-AID (Olympus EVIS X1) and CADEYE (Pentax) available in Colombia — GSMED
The two endoscopic AI systems with the most evidence available in Colombia — with demonstration and technical support.
Related articles
Sources
- ESGE. Position Statement on AI in Digestive Endoscopy. esge.com. 2023. ESGE 2023
- Hassan C et al. ENDO-AID CADe system — meta-analysis. Gastroenterology. 2024. ADR+10–15%. Hassan 2024
- van der Zander Q et al. CADEYE — pooled ADR meta-analysis. Endoscopy. 2024. ADR+12–15%. van der Zander 2024
- Repici A et al. CADe for colorectal polyp detection — multicenter RCT. Lancet Gastroenterol. 2024. Repici 2024
