Prediction of Lung Diseases based on Supervised and Unsupervised Approach

Abstract: Technology has played a vital role in health sector for past decades whether in detection of various diseases from early stage or in case of targeted treatment. In many Lung disease cases it's the same scenario the early stage detection of the diseases increases the survival rate of patient In this project the first step used in order to detect the lung disease based on unsupervised learning algorithm modelling is done by using LocNet and the optimized CNN is used for image dataset to compare. For the second step the images are differentiated based on whether lung tumor is either malignant or benign using supervised learning algorithm. In the project here two sets of images with data are considered to estimate the diseases where one set of data contains around 5,606 images whose size 1024 x 1024 pixel with height and width there are 15 classes and another set of data and images is the "full-dataset" that as 112,120 total images whose size 1024 x 1024 pixel with height and width.

Keywords: Optimized CNN, Supervised learning, unsupervised learning, Lung diseases