Algorithm 2. The ResNet50 model description.

Input: CXR images of dimension 500 × 500 pixels from the Training CXR-Dataset Output: ResNet model weights 1. epochs ← 100 2. for each image in the dataset do 3. resize image to 224 × 224 pixels 4. normalize the image pixels values from (0,255) to (0,1) 5. end 6. Load the ResNet50 model pre-trained on the ImageNet dataset 7. Make non-trainable all the layers of the model 8. Add a Flatten layer to the model output to obtain 1-D array of features 9. Apply a batch normalization to the 1-D array of features 10. Add a fully connected layer with 256 hidden neurons 11. Apply a dropout for inactivate units (50%) in the previous layer 12. Add a fully connected layer with 128 hidden neurons 13. Apply a dropout for inactivate units (50%) in the previous layer 14. Apply a batch normalization 15. Add a fully connected layer with four hidden units and a softmax activation function. 16. Optimize the model with Adam optimizer using a learning_rate = 0.0001 and a decay = learning_rate/epochs 17. Train the model for the given number of epochs and a batch size of 32 18. Save the final model