Diagnosis of Acute Respiratory Syndromes from X-Rays using Customised CNN Architecture
Palaniappan Sambandam, S. Varshaa Sai Sripriya, Amalladinna Rama Lalitha Pranathi, M. Muthulakshmi
Abstract
This work presents the diagnosis of various acute respiratory syndromes using customized CNN architecture from X-ray images. Complications of viral pneumonia results in influenza and COVID-19. The respiratory syndromes occur due to bacterial and fungal infections as well. Hence, the objective was to use customized CNN architecture to perform a multi-class pneumonia classification. VGG16 architecture is carefully trained for pneumonia classification with ReLU activation and categorical cross-entropy loss function. The proposed model is efficient and robust and yielded 97.87% accuracy on the train set and 90% accuracy on the test set. The experimental results suggest that the model efficiently detects all sorts of lung diseases, including COVID 19.