An Automated CHNN Model for the Classification and Detection of Lung Diseases using Transfer Learning

The COVID-19 pandemic began in December2019 and caused a global crisis. The WHO declared it a pandemic on March 11, 2020. Since October 10, 2020, COVID-19 has affected 200+ countries, causing over 37 million confirmed cases and 1 million deaths. RT-PCR is the usual method for detecting it, but it has drawbacks. Individuals who exhibit symptoms of COVID-19 but receive negative results from RT-PCR tests may be diagnosed with the disease using chest X-rays and CT scans, as these imaging techniques are capable of detecting lung abnormalities that are commonly associated with COVID-19, including consolidation and ground-glass opacities. The detection of COVID-19 systems faces numerous challenges, including false negatives, limited testing capacity, a scarcity of imaging equipment, and a shortage of data. With the increasing number of cases, there is a pressing need for a quicker, more cost-effective screening method. Chest X-ray scans can serve as a supplementary or confirming approach as they are fast and readily available. An Automated Hybrid Convolutional Neural network-Hopfield Neural Network (CHNN) is proposed in this study by extracting the features using VGG-19 for the classification and detection of lung diseases. In this work, both two-fold and multi-class classifications have been done with 99% and 97% accuracy respectively.