Thoracic Diseases Prediction Algorithm From Chest X-Ray Images Using Machine Learning Techniques

Abstract

Examining Chest X-Ray (CXR) is a time consuming process. In some cases, medical experts had overlooked the diseases in their first examinations on CXR, and when the images were reexamined, the disease signs are detected. Radiologists have to spend time diagnosing these chest X-ray images to find any potential lung diseases. Diagnosing X-ray require careful observation and knowledge of anatomical, physiology and pathological principles. The work involves machine learning techniques applied for automated prediction of seven thoracic diseases namely Pneumonia, Fibrosis, Hernia, Edema, Emphysema, Cardiomegaly and Pneumothorax from chest X-ray images. Computerized image segmentation and feature analysis helps in assisting the doctors in treatment and diagnosis of diseases more accurately.

Keywords: Thoracic Diseases, Independent Binary Classifier, SIFT (Scale Invariant Feature Transform), Visual Bag of Words, Logistic Regression, SVM (Support Vector Machines)