Pneumonia Detection Using Convolutional Neural Networks
Shanay Shah, Heeket Mehta, Pankaj Sonawane
Abstract
Pneumonia is a fatal disease that majorly affects the elderly and may sometimes prove to be life threatening. Early diagnosis of Pneumonia gains a paramount importance for saving many human lives. This paper aims at the detection and classification of patients affected by Pneumonia based on their chest X-rays. A convolutional neural network is employed from scratch to make the above diagnosis and yield highly accurate results. Deep Learning models automate the process and ensure speedy, adroit, and adept results when provided with X-rays of patients. The classification occurs after the image is fed through a series of convolutional and max pooling layers that are activated by using the ReLU activation function that is subsequently fed into the neurons present in the dense layers and finally, the output neuron is activated by the sigmoidal function. The accuracy increases as the model trains and decreases the loss simultaneously. Overfitting is prevented by implementing data augmentation before fitting the model. Thus, efficient, and cogent results are obtained by the proposed deep learning models to classify the chest X-rays for the detection of pneumonia.