A review on deep learning based pneumonia detection systems

Abstract

Pneumonia is a disease in which small air sacs of lungs called alveoli are affected. Pneumonia is a life- threatening disease caused by bacteria, fungi, or viruses. Millions of children die every year due to pneumonia. Earlier, only highly experienced radiologists examined pneumonia, but recent development in Image Processing and Deep Learning automated detection of diseases is possible. Early detection of pneumonia can save many lives. Various data augmentation techniques make the model achieve high accuracy even without having large datasets. Multiple types of research have used different techniques like CNNs, Transfer Learning, and Ensemble Learning. One of the current disadvantages of pneumonia detection is not having high-annotated data sets and the patient's medical history. The paper presents the overview of literature currently available on pneumonia classification by using chest x-ray. After reviewing the topic, the review investigates a comparison of various automated systems' drawbacks and limitations.