LUNG DISEASE DETECTION FROM X-RAYS USING CNN

Abstract

Lung diseases are becoming more commonplace throughout the world. Some of the major diseases include chronic obstructive pulmonary disease, pneumonia, asthma, tuberculosis, fibrosis, etc. A multitude of distinct image processing and
machine learning models have been developed for this cause. Various types of deep learning methods including convolutional neural network (CNN), vanilla neural network, capsule network and visual geometry group based neural network (VGG) have been implemented for lung disease diagnosis. For implementation of the research, Jupyter Notebook, Tensorflow, OpenCV, and Keras are utilized. The model is applied to NIH chest x-ray image dataset obtained from the Kaggle repository. Complete and sample editions of the dataset are kept in view. For the use of full dataset, CNN exhibits a validation accuracy of 90%. Whereas the use of sample dataset yields a much lower training time at the cost of a slightly less validation accuracy. Thus, the proposed CNN framework will make the diagnosis of lung diseases an easy task for medical practitioners as well as for experts.