Development of CheXNet-Based Web Application to Detect Pneumonia Using Chest X-Ray Images

Pneumonia is a severe respiratory infection that primarily affects the lungs, causing inflammation in the air sacs. an inflammatory condition affecting the lungs, poses significant health risks worldwide, especially in vulnerable populations like the elderly and young children. Early identification and accurate diagnosis are critical to timely treatment and positive patient outcomes. While, identifying pneumonia can be difficult and necessitates the expertise of a radiologist with considerable training. CNNs, one type of DL model, have the prospective to help identify and diagnose pneumonia earlier by classifying it. The chest x-ray dataset openly available in kaggle, which is the source of many high-quality x-ray pictures of different lungs of patients was used in this work. This study proposed CheXNet model which implements remarkably well in the categorization of pneumonia. It has numerous dense, pooling, and convolutional layers. Oversampling the training data is done to resolve the imbalance of classes. The model was compared to a number of ML classification techniques, including Random Forest, DBN, CNN, KNN, and SVM, as well as feature extraction models, including DensNet, ResNet50, VGG-16, Inception ResNet and CheXNet. The model learns the complex patterns and features seen in the images by training for 60 epochs with a batch-size of 128. This improves classification performance. With an accuracy of our suggested model is 98.4%. We test our work through a web application written in Python using the Flask framework which accepts the input image and predicts for normal or pneumonia.