Manuscript received November 13, 2023; revised December 27, 2023; accepted February 18, 2024; published August 21, 2024
Abstract—Pneumonia is an inflammation of the lungs which
is caused by bacteria, viruses, mold, and less commonly by
environmental toxins. Pneumonia is extremely prevalent
worldwide and is the number one cause of death among children
under the age of five and is the most common reason for
hospitalization for adults. Chest X-rays are a common medical
tool for diagnosing this illness, but must be analyzed by trained
radiologists, which is often time consuming and expensive.
Therefore, it would be beneficial to have an accurate automated
system for diagnosing pneumonia from radiological diagnostic
imaging. A variety of machine learning techniques have been
applied to the problem of medical image diagnostics and have
exceeded the accuracy of the average radiologist. Medical
datasets often suffer from a sparsity of training examples, so
data augmentation is often necessary. Here we implement an
auxiliary classifier generative adversarial network method
which generates synthetic X-ray images which augment the
training of a discriminator network. The described method has
an accuracy of 97.7% when trained on the Pneumonia MNIST
dataset, which is composed of low-resolution pediatric chest Xrays.
Given the common difficulty of acquiring significantly
sized medical datasets, the network was trained on a range of
datasets sizes to determine the impact on performance given a
smaller population of pneumonia examples. Even when trained
on a subsample containing only 20 examples, the network
achieves an impressive 84.54% accuracy. This system could be
used in areas lacking proper medical personnel or act as a
verification tool for diagnosticians.
Keywords—auxiliary classifier generative adversarial
network, Auxiliary Classifier Generative Adversarial Network
(AC-GAN), neural network, pneumonia, X-ray, pediatric,
machine learning, medical diagnostics.
Cite: Robert Langenderfer, "Auxiliary Classifier Generative Antagonist Network for the Detection of Pneumonia," International Journal of Machine Learning vol. 14, no. 3, pp. 77-83, 2024.
Copyright © 2024 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).