Deep Learning Scheme to Classify Chest X-ray into Pneumonia/Covid19: A Study
A. S. Vickram, Mathan Muthu C M
Abstract
Lung is a major organ in the respiratory system and the disease in lung will cause various health issues and lead to severe berating issues, when left untreated. The lung infection due to pneumonia/Covid19 will lead to harsh health problem if left untreated. Medical imaging based detection and treatment of lung illness is commonly executed in the hospitals and the chest radiograph (X-ray) is a common modality for lung abnormality screening. This research aims to implement the Deep-Learning (DL) method to detect the lung abnormality using X-ray images. The different sections in the proposed method includes; image collection and converting it to 224x224 pixels, feature extraction using chosen DL-model, feature optimization using Elephant-Herd-Algorithm (EHA) and fused-features generation, and binary classification with 5-fold cross validation. This work considered the ResNet-variant model to examine the X-ray and the achieved result confirms that this scheme presents a detection accuracy up to 98% when SoftMax based classification is executed.