Early Stage Emphysema Detection in Chest X-ray Images: A Machine Learning based Approach

Abstract

The objective of this study is to develop and evaluate a machine learning model for identifying emphysema in chest X-ray images using the NIH chest X-ray dataset at early stage. Emphysema being one of the world’s deadliest disease according to WHO has no cure and can only be prevented or detected. It’s symptoms being the same as the regular cough-cold makes it difficult to be identified and patients reach the last stage from where coming back is not possible. It is important that it gets detected at the earliest stage of its happening i.e Cardiomegaly. Chest x-rays are widely used to detect emphysema, and machine learning models have shown promising results in improving the accuracy and efficacy of this approach. Using Binary classification for ‘Cardiomegaly’ and ‘Effusion’ to predict early stages of emphysema detection by filtering the X-ray data set for these two classes namely ‘Cardiomegaly’ and ‘Effusion’. A part of the dataset was utilized for training the model, while a different test set was used for its assessment.The study’s findings showed that the binary classification using the SVM model achieved an accuracy of 79.95% on the test set, demonstrating its ability to effectively classify the two classes.

Topics & Concepts