Detection of Pneumonia from Chest X-Ray Images Using Convolutional Neural Network (CNN)

The bacterial or viral infection that causes pneumo-nia results in fluids filling the lungs, making breathing difficult. Early treatment or diagnosis is very important for this disease. It helps to increase the survival rate of patients. Every year, millions of individuals die because of Pneumonia. Even though Pneumonia causes deaths all throughout the world, underdeveloped countries are more affected than industrialized ones. Chest X-ray images are the key factor for this disease's diagnosis. Doctors examine the chest X-ray images to find pneumonia. However, any doc-tor would find it challenging and time-consuming to diagnose pneumonia using X-ray images.Even conventional pneumonia treatment is expensive for the majority of people in developing nations. We created a deep learning model (CNN) that can recognize pneumonia from the chest X-ray images. To achieve this, we proceeded in four steps: 1. Data preprocessing to shape and create enough data for our algorithms. 2. Neural architecture search to find the best-performing architecture. A small network architecture managed 91 % accuracy. 3. Hyperparameter tuning to tweak the hyperparameters. It resulted in the same network hyperparameters as from the neural architecture search. 4. Variable testing to find that a high learning rate, only ten epochs and ‘adam’ optimizers gave the best-performing network with an accuracy of 93 %. Despite the low resolution of the dataset, this result shows excellent potential for the classification task of pneumonia in images.