STM Journals Peer-reviewed research from Consortium e-Learning Network

Journal of Artificial Intelligence Research & Advances

Diagnosing Pneumonia from Chest X-rays Using Deep Learning Algorithms Through Convolutional Neural Network, Transfer Learning and Fine Tuning

  1. Kinjal Goswami
  2. M. Bhanu Sridhar
  3. Syed Suhana

Abstract

Pneumonia is an inflammatory condition of the lungs that induces air sacs which leads to a contagious infection of lungs. Patients who are afflicted with the virus can be saved from death and the virus can be eradicated from spreading further through effective diagnosis. X-rays of the chest are frequently used to diagnose pneumonia. Detecting pneumonia from a Chest X-ray is typically slow and inaccurate. It is essential to identify pneumonia quickly so that patients can receive prompt care, especially in rural areas. This work proposes a system that evaluates chest X-rays and categorizes the images using Deep Convolutional Neural Network Architecture, Transfer Learning and Fine Tuning on different CNN architectures. As part of this project, the implemented algorithms include CNN, VGG16, and Xception. Prior to building a model, Data Augmentation and Data Balancing is performed in order to improve the model’s generalization performance and accuracy. Among the above-mentioned algorithms, it has been deduced that VGG16 model has returned best accuracy.

Keywords

Support