Computer-aided detection for esophageal achalasia (with video)

Area of Science:

• Gastroenterology
• Medical Imaging
• Artificial Intelligence

Background:

• Achalasia is an esophageal motility disorder often missed during endoscopy, impacting patient quality of life.
• Computer-aided detection (CAD) software shows promise for achalasia diagnosis but requires clinical validation.

Purpose of the Study:

• To evaluate the clinical benefit of a novel computer-aided detection (CAD) software in diagnosing achalasia during endoscopy.
• To compare the diagnostic performance of endoscopists with and without CAD assistance.

Main Methods:

• 83 endoscopists from 27 centers assessed 50 endoscopic videos (25 achalasia, 25 nonachalasia) with and without CAD.
• The primary endpoint was improved sensitivity for non-experienced endoscopists.
• Sensitivity, specificity, and accuracy were compared using the McNemar test.

Main Results:

• Overall sensitivity for achalasia detection increased from 74.2% to 91.2% with CAD (17.1% improvement).
• Sensitivity significantly improved for non-experienced endoscopists (66.9% to 91.9%) and experienced endoscopists (79.5% to 90.8%).
• Accuracy and specificity also showed significant improvement with CAD assistance for all readers.

Conclusions:

• CAD software significantly enhances achalasia detection in a clinical setting, validating preclinical findings.
• The diagnostic benefit of CAD is more pronounced in endoscopists with less experience in achalasia.
• CAD implementation can facilitate timely treatment by reducing false-negative diagnoses.