Neural network architecture for differentiating Covid19 and viral pneumonia

Abstract

Covid-19 has wreaked havoc on the world when in some countries had cases in ten thousand each day thus, leading to a load on the healthcare system. Meaning that doctors and nurses had to spend more time on diagnostics. Therefore, one of the methods for reducing this load was to use a neural network for differentiating between covid and pneumonia cases. This citation showcase how neural networks can be used to detect lung x-rays having covid and pneumonia. Recall, precision, and f1-score measures are utilized to optimize the adaptive brightness of the images, selection process, resizing, and tune the neural network architecture parameters or hyperparameters. Classification quality metrics values over 91% depicted a decisive difference between radiographic images of patients having COVID-19 and pneumonia. Making it possible to make a model with strong forecasting capacity without pre-training on data from the 3rd party or engaging ready-to-use complicated neural network models. It can be the next step for the advancement of reliable and sensitive COVID-19 diagnostics.

Ähnliche Arbeiten