This project will use deep learning to classify colonoscopy images of different severity of ulcerative colitis, so as to assist clinicians in the accurate diagnosis of ulcerative colitis.

In this project, artificial intelligence was used to colonoscopic images of patients with ulcerative colitis with different disease activity levels and classify them according to the evaluation standard Mayo score to assist endoscopists in identifying disease activity levels of patients with ulcerative colitis during colonoscopy. It can help clinical endoscopists to accurately identify, and the visualization technology of artificial intelligence category response map can comprehensively display the areas with high importance for deep network classification results, and visualize the experimental lesion sites, thus effectively verifying the reliability and interpretability of deep network. This study can provide strong support for accurate identification of disease activity in clinical ulcerative colitis, effectively reduce the workload of clinicians, and provide a convenient, effective and practical clinical teaching tool.

Inclusion Criteria:

1. Subjects were 18-72 years old, male and female;
2. Clinical diagnosis of ulcerative colitis;
3. The subjects underwent colonoscopy and the colonoscopy report was complete.

Exclusion Criteria:

1. Subjects are younger than 18 years old or older than 72 years old;
2. Subjects underwent colectomy, ileostomy, colostomy, ileostomy, or other intestinal resection;
3. subjects with ambiguous diagnosis.

The images were retrieved from the endoscopy database of Army Medical Center of PLA and obtained from the colonoscopy of patients with ulcerative colitis who met the inclusion criteria. The data images were used to establish and verify the model of ai-assisted recognition system.