Artificial Intelligence in Inflammatory Bowel Disease Endoscopy: Advanced Development and New Horizons
Table 1
Endoscopic AI applications in diagnosing UC.
Author
Year
Data source
Purpose
Results
Sutton et al.
2022
851 images of UC patients
To differentiate UC from other intestinal diseases and to evaluate the severity of UC endoscopic ulcers
The accuracy (87.50%) and area under the curve (AUC, 0.90)
Takenaka et al.
2020
40,758 images of endoscopies and 6,885 biopsy outcomes of 2012 UC patients
To create a deep neural network system for analyzing endoscopic images of UC patients
The remission in endoscopy with 90.1% accuracy (95% and 89.2–90.9%)
Najarian et al.
2021
Video of the clinical trial set with 51 high resolution and 264 tests
To trial a fully automated video system for analyzing and grading endoscopic disease in UC
Automated Mayo endoscopic subscores (MES) scoring of clinical trial videos correctly differentiated between remission and active disease in 83.7% of cases
Gottlieb et al.
2021
795 full-length endoscopy videos of 249 patients
To verify a deep learning, algorithm can be trained to predict levels of UC severity from full-length endoscopy videos
(0.787–0.901) for endoscopic Mayo Score (eMS) and 0.855 (95% confidence interval, 0.80–0.91) for UCEIS
Bossuyt et al.
2020
29 consecutive patients with UC and 6 healthy controls
To develop an operator- independent computer-based tool to determine UC activity based on endoscopic images
RD correlated with Robarts histological index (RHI) (,), MES (,), and UC endoscopic index of severity scores (,)