Application of deep learning techniques for detection of COVID-19 cases

The New Coronavirus Disease (COVID-19) is a new virus that emerged as a result of research in a group of patients who showed respiratory tract infection symptoms such as fever, cough, and shortness of breath. It is very important to prevent the spread of the disease with rapid diagnosis methods and to detect positive cases early. The most common testing technique used for diagnosing COVID-19 is RT-PCR, which is a real-time reverse testing technique. However, the long duration of pathological laboratory tests and inaccurate test results have led researchers to different fields. Radiological imaging has begun to be used to monitor COVID-19 disease as well as being useful in detecting various lung diseases. The application of deep learning techniques together with radiological imaging is very important in the correct detection of this disease. In this study, the effect of basic fusion functions on the classification performance of ensemble learning algorithms was investigated using the COVID-19 X-Ray dataset. Two different ensemble models were created to combine different deep learning models; ENS-1 and ENS-2. In these ensemble models, basic fusion functions such as Max, Mode, Sum, Mean and Product have been tested. When the obtained values are examined, it is seen that the Max and Product basic fusion functions have a positive effect on the classification performance. In multiclassification, the Max function for both ENS-1 and ENS-2 stands out with %85 and %86 accuracy, respectively. Product function achieved the highest performance with %99 in binary classification. The results show that fusion methods can achieve better classification performance in binary classification.