This is a project to classify four classes of pancreatic cancer grade from a pathology image using deep learning and CNN. This repository contains the source codes used for this project. The source codes for training deep learning models and also for the back-end of the web appliction were developed using Google Colab, a Google's version of Jupyter Notebook. The GUI for the web application was developed using Anvil. All source codes were written in Python language.
Sourcecode was developed and tested with:
- Python - 3.6.9
- Keras - 2.3.0
- TensorFlow - 2.3.0
| Source code | Descriptions |
|---|---|
| Colab - Training | Algorithms for transfer learning, model development, data augmentation, training and evaluating deep learning models. |
| Colab - Web Application (Back-end) | Algorithms for fetching and sending image from Anvil application via Anvil Uplink. Also contains algorithm used for slicing and stitching image, and making prediction using a deep learning model. |
| Anvil - Web Application (GUI) | Python code that defines the layout of the web application. |
- Open Anvil Web App GUI.
- Load System_web_app_(Backend).ipynb python notebook into Google Colab.
- In Colab, goto "Load Model" section. Edit the model_directory to the directory of the the model in Google Drive (/content/gdrive/MyDrive/...). Rename the model_name to the file name of your model.
- In Anvil, goto setting(gear icon) > Uplink. Copy the uplink key.
- In Colab, goto "Main" section. Make sure the string of uplink_key is similar to the one in Anvil. If not paste the new key.
- In Anvil, click run.
- In Colab, run the "Library", "Load Model" and "Main" sections. The "Main" section will run forever waiting for interaction in Anvil.
- DONE!