Detection of Pneumonia using Convolutional Neural Network

Pneumonia accounts for over 15% of all deaths of children under 5 years old internationally. In 2015, 920,000 children under the age of 5 died from the disease. Pneumonia is one of the leading causes of death in developing countries like India too. While common, accurately diagnosing pneumonia is a tall order. The general procedure will require highly qualified doctors to diagnose chest x-rays and confirmation through the patients clinical records, vital signs and laboratory exams. However, there are many other reasons which makes the diagnosis of pneumonia very complicated such as bleeding, fluid overload, volume loss, lung cancer, or post-radiation or surgical changes. Other factor to detect pneumonia is the fluid in the pleural space which appears as increased opacity on the lung x-rays. By making the comparison of the x-rays taken at different time points and identifying the correlation clinical symptoms and records are helpful for the diagnosis of pneumonia. X-rays are the most commonly performed diagnostic imaging study. The other factors like depth of inspiration and the positioning of the patient can change the appearance of the x-rays, complicating interpretation further. In developing countries like India there is a lack of infrastructure and medical experts in rural areas to provide the diagnosis of such diseases. Thus there is a necessity of an automated system that can detect pneumonia.