Pneumonia Detection in Chest X-rays using CNN

Pneumonia affects millions globally and requires prompt and precise diagnosis. Chest X-rays are commonly used but manual analysis is time-consuming and prone to error. We developed an automated method using CNNs for pneumonia identification through chest X-ray interpretation. We utilized a large dataset comprising thousands of chest X-ray images from both pneumonia-positive and pneumonia-negative patients. By creating, training, and analyzing CNN models, we were able to detect subtle patterns and features indicative of pneumonia. To optimize the performance of our models, we tested multiple CNN architectures such as the baseline CNN, CNN using Keras, transfer learning, and transfer learning using FT, on the same dataset[1], using evaluation criteria such as testing accuracy, training accuracy, and Validation accuracy. Our results demonstrated that the CNN-based technique outperforms standard computer-aided diagnostic (CAD) systems and human radiologists in terms of detection accuracy. Our method achieved an accuracy of 73%, 84%, and 92.1% The model Transfer learning using ft showed the highest accuracy of 92.9 %. Our deep-learning CNN models enable accurate detection of pneumonia, even from low-quality, low-dose X-ray images, improving patient care.