Using Deep Learning to Classify Ulcerative Colitis and Colorectal Cancer Images
Author Information
Author(s): Carreras Joaquim, Roncador Giovanna, Hamoudi Rifat
Primary Institution: Department of Pathology, School of Medicine, Tokai University
Hypothesis
Can deep learning effectively classify images of ulcerative colitis, colorectal cancer, and normal colon tissue?
Conclusion
The convolutional neural network classified the three diagnoses with high performance, and LAIR1 and TOX2 were found to correlate with the severity of ulcerative colitis.
Supporting Evidence
- The CNN achieved 99.1% accuracy for ulcerative colitis, 99.8% for colorectal cancer, and 99.1% for colon control.
- LAIR1 expression was higher in steroid-requiring ulcerative colitis compared to mesalazine-responsive cases.
- TOX2 expression was lower in steroid-requiring ulcerative colitis compared to mesalazine-responsive cases.
Takeaway
Researchers used a computer program to look at pictures of the intestines and tell if they showed a disease called ulcerative colitis or cancer. The program did a really good job!
Methodology
A convolutional neural network was trained on histological images to classify ulcerative colitis, colorectal cancer, and normal colon tissue.
Limitations
The study was limited by the number of samples used for CNN analysis.
Participant Demographics
Mean age of participants was 38.4 years, with a male/female ratio of 20/35.
Statistical Information
P-Value
0.001
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website