DESIGN A CHEST X-RAY-BASED DIAGNOSIS PROGRAM

Abstract

The condition, components, and adjacent structures of the chest can be assessed via Chest X-Ray (CXR) image–based diagnosis. However, relying on CXR images to observe and make accurate diagnoses will take a lot of time, even for an experienced doctor. In addition, in order to minimize errors in disease diagnosis due to human factors such as doctors’ fatigue when looking at too many CXR images in a day, creating a smart tool that helps doctors reduce the time for CXR image inspection and diagnosis with high accuracy will save a lot of cost and time. This article will study the advanced algorithms of neural networks in deep learning and apply these algorithms to a big data set of CXR images to design deep learning models on several common diseases. The proposed models help users to diagnose their own CXR images or doctors can have a quick and mass diagnosis channel from CXR images with high accuracy through a diagnosis program based on Web Explorer.