Evaluation of ulcerative colitis with deep neural networks based on endoscopic images

UMIN ID: UMIN000031430

Registered date:15/03/2018

Basic Information

• Recruitment status: Complete: follow-up complete
• Health condition(s) or Problem(s) studied: Ulcerative colitis
• Date of first enrollment: 2018/04/01
• Target sample size: 500
• Countries of recruitment: Japan
• Study type: Observational

Outcome(s)

• Primary Outcome: accuracy of DNN-UC to evaluate endoscopic and histological healing
• Secondary Outcome: i) ability of DNN-UC to score UCEIS ii) accuracy of DNN-UC for endoscopic and histological healing in each segment iii) accuracy of DNN-UC in each confidence case iv) accuracy of DNN-UC stratified with the degree of colon cleaning.

Key inclusion & exclusion criteria

• Age minimum: 15years-old
• Age maximum: Not applicable
• Gender: Male and Female
• Exclude criteria: i) patients with prior colon surgery, IBD unclassified, Crohns disease, colorectal neoplasia, or concomitant infectious colitis ii) patients for whom colonoscopy were contraindicated iii) patients for whom biopsy were contraindicated because of blood disease or antithrombotic or anticoagulation therapy.

Related Information

• Primary Sponsor: Department of Endoscopy, Tokyo Medical and Dental University
• Source(s) of Monetary Support: the Cooperation Program between Tokyo Medical and Dental University and Sony Imaging Products & Solutions Inc.