Healthy Individual
Conditions
Keywords
Convolutional neural network, Deep learning, Endoscopy, Gastric neoplasms
Brief summary
This study aimed to find the CDSS-assisted upper GI endoscopy has clinical benefit compared to conventional routine screening upper GI endoscopy. Investigators will conduct a pilot randomized controlled study. Consecutive patients will be allocated either to CDSS-assisted upper GI endoscopy or conventional routine screening upper GI endoscopy. The lesion detection rate will be compared between both groups.
Detailed description
Artificial intelligence has been adopted in the field of gastrointestinal endoscopy. Investigators previously established deep-learning models to predict the histology and invasion depth of gastric lesions using endoscopic stillcut images. However, clinical benefit of this model has not been evaluated. This study aimed to find the CDSS-assisted upper GI endoscopy has clinical benefit compared to conventional routine screening upper GI endoscopy. Investigators will conduct a pilot randomized controlled study. Consecutive patients who visited Chuncheon Sacred Heart hospital will be allocated either to CDSS-assisted upper GI endoscopy or conventional routine screening upper GI endoscopy. The lesion detection rate will be compared between both groups. Expert endoscopists will conduct this randomized study.
Interventions
OTHERCDSS-assisted upper GI endoscopy
Expert endoscopists perform screening upper gastrointestinal endoscopy while looking at the monitor (CDSS system: automatic detection or lesion classification function) and listening to the sound of the monitor next to the original endoscopy monitor system.
OTHERno intervention
Expert endoscopists perform routine screening upper gastrointestinal endoscopy without CDSS's help.
Sponsors
Chuncheon Sacred Heart Hospital
Study design
Allocation
RANDOMIZED
Intervention model
PARALLEL
Primary purpose
OTHER
Masking
NONE
Eligibility
Sex/Gender
ALL
Age
20 Years to No maximum
Healthy volunteers
Yes
Inclusion criteria
* Participants who visited for routine screening upper gastrointestinal endoscopy
Exclusion criteria
* Participants who refuse to participate in this study * Participants who showed paradoxical reaction prohibiting routine examination
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Gastric lesion detection rate | Baseline (gold standard; ground truth is expert endoscopists' lesion detection) | Gastric lesions (low or high grade dysplasia, early gastric cancer, or advanced gastric cancer) detection rate |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Lesion classification accuracy | Baseline (gold standard; ground truth is histologic diagnosis of detected lesions) | Whether CDSS correctly diagnosed detected lesions or not |
Countries
South Korea
Contacts
Primary ContactChang Seok Bang, MD., PhD.
Outcome results
None listed