The aim of this tutorial is to classify patches of colorectal tissues in 9 different classes [1] using transfer learning.
We will run 3 experiments in parallel in order to better understand the topic:
- Using transfer learning & retraining all layers
- Using transfer learning & freezing half of the layers
- Training from scratch (= no transfer learning)
[1] J. N. Kather et al. (2018). 100,000 histological images of human colorectal cancer and healthy tissue (Version v0.1) [Data set]. Zenodo. http://doi.org/10.5281/zenodo.1214456
The tutorial is implemented in tensorflow 2.0 in a jupyter notebook: transfer_learning_classification.ipynb
- Jupyter notebook
- Python (tested with: 3.7)
- Tensorflow (tested with: 2.0)
My name is Marie Duc, I am Research Scientist at the Werner Siemens Imaging Center in Tübingen Germany. I am working on applying machine learning to medical imaging.
By any questions, requests, or anything else, please contact me!! :)
This project is MIT licensed.
This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 764458.