Mini-Grant Award: A Rest API to Classify Pneumonia Infection From Chest X-ray Images Using Machine Learning Techniques and Linked to a Mobile Application

Featured Image: Bacterial pneumonia (Source: Wikipedia)

According to WHO statistics published in 2020, the number of people killed by influenza and pneumonia in South Africa reached 39,341, accounting for 8.65% of all fatalities. Thus, worldwide, South Africa is ranked at number 40 with an age-adjusted death rate of 83.23 per 100,000 inhabitants. In instantaneous, pneumonia is among the ten leading underlying natural causes of death in South Africa. The situation is even worse in rural areas because of several reasons, among them; not having enough X-ray machines, few expert doctors (radiologists) in rural areas to analyse and interpret the X-ray pictures to determine if the pictures are normal pictures or if patient has pneumonia caused by bacteria or viruses.

It is important to note that pneumonia is a curable disease if it is identified early. The ability to accurately classify these two types of pneumonia (viral or bacteria) can guarantee effective treatment which will boost survival chances. Therefore, there is a need to find a cost-effective and efficient method to easily classify, analyse and interpret X-ray images.

Therefore, this study seeks to:

  • Create a simple model that would classify whether a person has pneumonia-bacteria, pneumonia-virus, or normal given the chest-x-ray image. The model should run on a mobile phone without requiring more computational power (backed with cloud).
  • Build a model that converts the English interpretations into South African local languages like isiXhosa, Zulu, Venda, and many others. Thus, making it easier for the local communities to understand and accept the use of technology as they will have a sense of belonging.


  1. Dr William Vambe, Walter Sisulu University
  2. Lenoah Lindiwe Baloyi, Tshwane University of Technology

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top