Detecting Pneumonia from X-Ray Images using Transfer Learning

Pneumonia is one such deadly infectious disease that causes inflammation the air sacs and could be life threatening if not diagnosed and treated early. Diagnosis based on symptoms and examination of Chest X-ray by a doctor to find the infection in lungs and severity of the Pneumonia, which is tedious and prone to errors. The main objective of this paper is to propose a Deep Convolution Neural Network (CNN) model based on Inception-V3 architecture and perform various hyper parameter tuning to get better accuracy than original Inception-V3 to classify the X-Ray images and find whether the person has pneumonia or not. This paper presents a model which reduces the abruptness in the Inception-V3 model while converting to a binary classification model. The proposed network consists of 130,514 trainable parameters which is significantly less compared to models that are not pre-trained and do not use transfer learning. Data augmentation is performed to improve the veracity of the data and have evaluated the results in varying environments by applying batch normalization and dropout separately and evaluated the accuracy.