Deep Learning Analysis of COVID-19 Lung Infections in CT Scans

Everyone observed the unbearable hardship of the coronavirus. The necessity of automatic coronavirus detection has been realized from the dire circumstances. Among the various diagnosis techniques lung CT scan image analysis has become a beneficial approach due to it's precise detection of lung abnormalities. In our research, we suggest a novel deep learning model for the segmentation of COVID-19 from CT scan images. Our suggested architecture is a complicated blend of the UNet and MobileNetV2 models which obtains remarkable accuracy in precise segmentation of COVID-19 affected lesions in lung CT scan images. In our study, we used advanced approaches such as picture reorientation to achieve ideal alignment, histogram analysis for thorough data assessment, and upsampling to determine our results. This rigorous approach allows for the incorporation of a wide range of masks, including lung and infection masks, as well as their combinations. Using many masks, the model gains a thorough understanding of complex lung architecture and infection patterns. The research significantly impacts COVID-19 image analysis because of its accurate outcomes and accelerates the advancement of computing technology in the application of the medical field. We achieved 95.56% accuracy on the 3D Unet model.