Predictive Analysis for Human Chest Diseases Detection Using Transfer Learning

People are being affected by pollution and irregular food habits day by day. This leads to the cause of many kinds of diseases, especially chest diseases such as Pneumothorax, Pneumonia, Effusion, Atelectasis, Nodule, Mass, Cardiomegaly, Edema, Lung Consolidation, Pleural Thickening, Infiltration, Fibrosis, and Emphysema. Machine learning, which is an important technique of artificial intelligence, has the ability to learn and predict the diseases automatically. The aim of deep learning in healthcare is its capability to extract the features from large number of datasets. Initially, pre-processing step is done to remove noise and to deal with blurred images from the database, which has training and testing datasets. Machine learning algorithms will learn the selected features from the training dataset and the gained experience will be stored as a separate data model that is recommended for predicting the chest diseases from the test dataset. The proposed method uses Convolutional Neural Network with inceptionV3 to predict the chest diseases. Transfer learning is used to extract the inceptionV3 features from the pre-trained model and the predictions are given in terms of a percentage. Thereby, the doctors can easily diagnose the diseases and save the lives of the patients.