The project includes working on a chest CT-scan series of images aiming to develop an automated classification algorithm solution for Covid-19/non-Covid-19 diagnosis. The dataset is called COV19-CT-DB obtained at AI-enabled Medical Image Analysis Workshop and Covid-19 Diagnosis Competition (MIA-COV19D), https://mlearn.lincoln.ac.uk/mia-cov19d/.
The codes develop CNN models as well as transfer learning models as follows:
-
The Code '3D_Image_Classification_binary_loss_function.ipynb' is a CNN model with binary crossentropy as a losss function.
-
The code 'Customized-CNN-Model-and-Tranfer-Learning-Model.py' is a customized CNN model with a linearly changing learning rate for image classification. The proposed model is compared to transfer learning models, MobileNet, for feature exctraction followed by a random forest classifier to take the final decision.
-
The code 'COV19-CT-DB-CNN-Model.py' Proposes a less hand-engineered CNN model Architecture for automated COVID-19 diagnosis
The CNN model achitechture is:
Dependencies:
-numpy == 1.19.5
-matplotlib == 3.3.4
-tensorflow == 2.5.0