An Efficient Convolutional Neural Network for Pneumonia Recognition from Chest X-Ray Images

A microbial infection that results in chronic inflammation of the lung cells causes pneumonia. It is mainly caused by bacteria, viruses, fungi or other microorganisms. It causes inflammation in the alveoli of the lungs. The patient with pneumonia faces difficulties breathing as their lungs become filled with pus or fluids. This can also cause chest pain, a fever, or coughing. Severe pneumonia is considered a primary cause of death among infants under the age of 5. Traditionally, the detection of pneumonia using chest X-ray images is entrusted to the ability of radiologists, which is quite time-consuming and prone to error. By identifying pneumonia cases at an early stage using an efficient CNN model, we can speed up the delivery of necessary medications following a precise diagnosis. The possible risk of complications can be reduced, improving patient outcomes. The disadvantage of the majority of the available diagnostic systems based on transfer learning models is that they demand a significant amount of computational power. Our proposed model is considered to be light-weight compared to other transfer learning models. We used a large dataset, which consists of a total of 12448 images. The model achieved an accuracy of 98% and the accuracy of our model surpassed that of some transfer learning benchmark models, including ResNet50v2, ResNet101V2, and ResNet152V2.