Deep Learning for Chest X-ray Analysis: Pneumonia Detection رسالة

Abstract

Pneumonia, which ranks among the primary causes of death on a global scale, poses a substantial threat to the health of children everywhere, especially those below the age of five. An accurate and prompt diagnosis is essential for effective therapy and patient outcomes. This study aims to analyze chest X-rays in the best method, which will provide a definitive method for pneumonia assessment. The investigation employed a comprehensive, global dataset acquired from the Kaggle platform. The dataset contains 5856 chest X-ray images. Data preprocessing is used in training deep learning models to improve the ability to predict pneumonia in X-ray images. After that, several deep learning models, namely ResNet, ResNet50, VGG16, and GoogLeNet, enhanced techniques based on the CNN model, have been used to detect pneumonia in X-rays. These models have been subjected to rigorous testing to guarantee their accuracy and performance. The results of the DenseNet techniques are highly effective, and the accuracy is very good. On the other hand, GoogLeNet and other techniques could not achieve the desired degree of accuracy. Considering the notable variations in performance, the DenseNet algorithm appears well-suited for this dataset type. Further research could investigate various algorithmic modifications to improve performance further. Additional examination of its composition and unique characteristics is required to ascertain its efficacy in this domain