ODE PathMNIST Classifier — Medical Pathology AI
By Llewellyn Systems Inc | Part of
ODE Science
Model Description
ResNet18-based classifier for colorectal polyp histology images. 9-class classification on PathMNIST benchmark.
Performance
- 95.96% validation accuracy on PathMNIST (9 classes, ~100K images)
- Trained on NVIDIA Tesla T4 GPU
- ResNet18 backbone with custom classification head
Classes
- Adipose | 2. Background | 3. Debris | 4. Lymphocytes | 5. Mucus
- Smooth Muscle | 7. Normal Colon Mucosa | 8. Cancer Stroma | 9. Colorectal Adenocarcinoma
About ODE Science
AI-powered scientific research platform featuring molecular modeling, medical imaging, protein analysis, and quantum computing.
