A group of researchers from Alanya Alaaddin Keykubat University has used deep learning and convolutional neural network to detect Coronavirus COVID-19 disease.
In their paper published at Physiol Genomics 2020, the group presented two powerful convolutional neural network (CNN) architectures for two different classification tasks in the domain of disease detection. The first model uses chest X-ray scans as input and performs a binary classification of whether the patient has the COVID-19 disease or not.
The second neural network takes the same type of input, a chest X-ray scan, but it classifies the scans into three exclusive categories: COVID-19, Normal, and Pneumonia. Researchers used publicly available X-ray datasets to train their proposed models. One such dataset is the Covid-19 dataset by Joseph Paul Cohen, Paul Morrison, and Lan Dao that contains 542 frontal chest X-ray images from 262 people. Another dataset used for the experiments is the Covid dataset created by Paul Money and it contains 5,863 X-Ray images and 2 categories: Pneumonia and Normal.
Researchers tuned the hyperparameters in an automatic manner using Grid search. Results from the experiments showed that the first neural network achieves 98.82 % accuracy, while the second one achieves 98.27 %.
More details about the architecture of the networks as well as the results from the experiments can be found in the paper.
