More than 1 million adults are hospitalized with pneumonia and around 50,000 die from the disease every year in the US alone (CDC, 2017). Tuberculosis (TB) remains the second leading cause of death due to infectious diseases worldwide1. The World Health Organization (WHO) has a Millennium
Development Goal aimed towards reversing and restricting the spread of the disease by 2015. World TB Day is held every year on 24th March, to raise awareness on the ongoing epidemic. For World TB Day 2015, the United Nations, the Stop TB Partnership and the WHO are urging all governments and health organizations to mobilize political and social commitment towards eliminating the disease as a public health burden. The World TB day theme this year is “Reach the 3 Million: Reach, Treat, Cure Everyone” - aimed at the three million people who are not treated every year.
Chest X-rays are currently the best available method for diagnosing pneumonia (WHO, 2001), playing a crucial role in clinical care (Franquet, 2001) and epidemiological studies (Cherian et al., 2005). However, detecting any abnormalities such as Atelectasis, Calcification, Cardiomegaly, Consolidation, Fibrosis, Nodule, Pleural Effusion, Pleural Thickening, Pneumothorax, and so on in chest X-rays is a challenging task that relies on the availability of expert radiologists. Those abnormalities are very crucial to determine theit relevant diseases like TB, pneumonia, and so on.
AXIR is used to detect digital chest images by being a component of a complete x-ray system. We developed a deep learning solution that can detect abnormalities from chest X-rays at a level comparable to Radiologist. Our algorithm, AXIR-Net, is a 571-layer convolutional neural network trained on our own Chest Image Dataset, currently the largest publicly available chest X-ray dataset, containing over 100,000 frontal view X-ray images with 14 diseases.