Rethinking Densely Connected Convolutional Networks for Diagnosing Infectious Diseases

Due to its high transmissibility, the COVID-19 pandemic has placed an unprecedented burden on healthcare systems worldwide. X-ray imaging of chest emerged as a valuable and cost-effective tool for detecting diagnosing patients. In this study, we developed deep learning model using transfer with optimized DenseNet-169 DenseNet-201 models three-class classification, utilizing Nadam optimizer. We modified traditional DenseNet architecture tuned hyperparameters improve model’s performance. The was evaluated novel dataset 3312 images from publicly available datasets, metrics such accuracy, recall, precision, F1-score, area under receiver operating characteristics curve. Our results showed impressive detection rate accuracy recall patients, 95.98% 96% achieved 96.18% 99% DenseNet-201. Unique layer configurations optimization algorithm enabled our achieve rates not only patients but also identifying normal pneumonia-affected ability detect lung problems early on, well low false-positive false-negative rates, suggest that it potential serve reliable diagnostic variety diseases.