Chest X-Ray Image Classification

Abstract

The chest X-ray abnormal and normal classification model will classify X-ray images of patients. The dataset is collected from hospitals for training. The state of art image classification algorithm (YOLOv5x-cls) was used to train the model. The model will classify the scanned Xray into normal and abnormal Chest X-rays. As there is a huge burden on a radiologist of 3Million X-rays per annum, our project will help them to get the report in a single click. The project is a breakthrough in radiology, due to the instant rise in diseases, doctors found it difficult to tackle them. So, our project will bring more ease to doctors, which will save time and help in accuracy doctors need to treat more patients in less time. The model is trained to achieve maximum Accuracy (83%) on the dataset. The trained model is used in a web app for online inference and readily results can be served to doctors in it.