Table 3.

Comparison about previous works on COVID-19 classification techniques.

Used technique	Description	Advantages	Disadvantages
Deep Convolutional Neural Network (DCNN) [2]	DCNN is a classification technique that was used for the detection of coronavirus infected patient based on three different convolutional Neural Network models.	Transfer learning allows the training of data with fewer datasets and requires less calculation costs.	One of the biggest limitations to transfer learning is the problem of negative transfer. Transfer learning only works if the initial and target problems are similar enough for the first round of training to be relevant.
Gray Level Size Zone Matrix with SVM (GLSZM-SVM) [17]	GLSZM-SVM is a hybrid method that extracted the features by using GLSZM and then used SVM to classify these extracted features.	SVM is a powerful classification method.	SVM was not suitable for large dataset and could not perform its task well when the dataset included more noise.
Deep Learning based Methodology (DLM) [18]	Deep features were extracted using pre-trained CNN and SVM was used for classification.	Pre-trained model is effective power in features extraction and SVM is a powerful classification method.	SVM was not suitable for large dataset and could not perform its task well when the dataset included more noise.
Deep Learning Algorithm (DLA) [19]	DLA has the ability to extract COVID-19’s specific graphical features to introduce a clinical diagnosis prior to pathogenic testing. This method tried to save critical time for the disease diagnosis.	It demonstrated the principle of using artificial intelligence to extract the radiological features for a timely and accurate diagnosis of COVID-19.	This method cannot provide the optimal accuracy.
Stack Hybrid Classification (SHC) method [20]	SHC is a COVID-19’s classification method that depended on ensemble methods that integrate several models for improving prediction performance.	The proposed SHC model is better than the traditional classification approaches to classify COVID-19 patients.	This method is slow.
Deep learning framework (COVIDX-Net) [21]	COVIDX-Net is the frame work included seven different architectures of deep convolutional neural network models to classify the patient status either negative or positive COVID-19 case.	Efficient to classify positive cases.	Cannot classify the normal cases correctly. Therefore, it requires another model in CAD systems to determine the health status of patients against COVID-19 in X-ray images.