Automated Pneumonia Identification Using Convolutional Neural Network

Pneumonia is a contagious disease usually caused by a bacterial infection in the alveoli of the lungs. When lung tissue is infected, pus forms in the infected area. Healthcare providers use physical examination and diagnostic procedures like chest X-rays, ultrasounds, or lung biopsies to diagnose whether a patient has pneumonia. Misdiagnosis or wrong treatment can drastically affect the quality of life for a patient. Advances in deep learning assist specialists to arrive at better conclusions concerning the diagnosis. This method utilizes an adaptive and light-weighted model of deep learning, such as a Convolutional Neural Network (CNN) for the identification and prediction regarding whether the patient is infected or not through his chest X-rays. The researchers used a database containing 20,000 images which were formatted at $224 \times 224$ pixels within batches of 32 images to check the model performance. The tested model demonstrated $90 \%$ accuracy in its performance tests. The study demonstrates its ability to both identify and forecast bacterial along with viral pneumonia from chest X-ray images through evaluation results.