Automated Diagnosis of Chest X-Ray for Early Detection of COVID-19 Disease

Algorithm 1

Deep-transfer learning-based COVID-19 classification.

Step 1. Input: the X-ray dataset for COVID-19, pneumonia, lung opacity, and normal.

Step 2. Apply a preprocessing process to enhance photos with the average filter.

Step 3. Divide the dataset into 80% for training and validation and 20% for testing.

Step 4. Apply the augmentation method to create balance in the dataset.

Step 5. Extract the most important representative features by convolution layers of two ResNet-50 and AlexNet models.

Step 6. Then, the transfer learning is applied to train the two models for the classification of COVID-19.

Step 7. The two models are trained based on enhanced hyperparameters.

Step 8. Evaluate the performance of the two models on the dataset through a set of measures (confusion matrix and ROC).

Step 9. Output: classify each image to one of the four classes (COVID-19, viral pneumonia, lung opacity, and normal).