Covid-19 Detection From CT Scan Images Using Deep Learning Models

Since 2020, the whole world has been dealing with a pandemic called coronavirus (COVID-19). Compared to automated methods, detecting the presence of the virus in the human body by manual methods is more expensive and time-consuming. Therefore detection of the virus by automated method has further stimulated our research interest. One of the most popular and important methods for automated detection of any disease is CT scanning of the lungs. Our study presents a novel technique that uses a state-of-the-art computer analysis to accurately classify and isolate lung infections by meticulously analyzing lung CT scan images. Our research uncovers minor differences among classes. So, it indicates the need for ongoing development through collaboration with medical specialists to address the minor differences. This study presents a comprehensive technique that uses sophisticated computer analysis to accurately classify and isolate lung infections from Lung CT scan images. The technique, which is a complex combination of some classical CNN architectures, performs admirably in terms of accuracy and is especially good at identifying spots of infection in lung pictures. Our study integrates advanced techniques such as EfficientNetB4, VGG19, VGG16, NASNetMobile, Xception and proposed a new model. The latters obtain an in-depth understanding of complex lung architecture and infection patterns by using several masks. We get 99.96% accuracy on our proposed model compared to other evaluation. This work highly benefits the field of COVID-19 image analysis and offers a thorough framework for real-world use in healthcare sectors. We also find that this proposed model is appropriate for the used dataset.