Predictive Accuracy Analysis for Corona Virus Using Residual U-Net in Radiological Data

Confirming a SARS-CoV-2 diagnosis requires proper assessment, and imaging techniques are vital tools for this purpose. Imaging techniques are crucial in this process, as they can assess the severity of the disease, detect any complications, and monitor the response to treatment. The lungs are the organ most affected by the virus, and chest X-rays are the preferred initial imaging test because they are easily accessible and cost-effective, whether acquired at regular X-ray facilities or using portable equipment. This research employed a dataset of people who had contracted the virus to identify and classify diseases. The classification can help doctors treat patients and save lives. To classify normal, COVID-19, and Viral Pneumonia classes, the researchers modified the Residual U-Net pre-trained network architecture, which proved to be an effective method. Compared to the previous research analysis, the residual U-Net provides a better result for classification. The researchers modified the Residual U-Net pre-trained network architecture to classify normal, COVID-19, and Viral Pneumonia classes and found that it provided better results compared to previous research. The accuracy of classification for this study achieved a 99.01% accuracy rate in disease classification.