Automatic Pneumonia Detection Through Chest X-Ray Image-Based

Pneumonia is a well-known inflammatory lung illness that needs to be diagnosed quickly and correctly to be effectively treated. Chest X-ray diagnosis of pneumonia is difficult, particularly in disadvantaged clinical settings. The confusing radiographic appearance of pneumonia frequently results in incorrect diagnoses and drawn-out diagnostic processes. Hence, this study aims to develop automated pneumonia detection prototype through X-ray image-based which can help in the early and efficient diagnosis of pneumonia, enabling timely treatment and potentially reducing the burden on radiologists or healthcare professionals. To accomplish this, Discrete Wavelet Transform (DWT) and bilateral filter are used to reduce noise from X-ray images. The classification was performed by using Convolutional Neural Networks (CNN) to generate evaluation metrics to determine the learning algorithm's performance. A set of 5,840 X-ray images of a normal chest and a pneumonia-infected chest were used for automation testing using CNN as models to generate the confusion matrix and performance accuracy. The trained CNN model shows encouraging results, categorising both pneumonia-positive and pneumonia-negative cases with outstanding accuracy, recall, and F1-score of 96.0%. This study shows the potential of automated pneumonia diagnosis utilising chest X-ray images, providing a helpful tool to help medical practitioners to make prompt and accurate assessments. The main drawback would be the absence of demographic information. Future suggestions include incorporating demographic data even more and enhancing the model's capacity to consider patients' histories. This study contributes to the development of automated pneumonia diagnosis by utilising deep learning techniques and improving diagnostic precision and effectiveness in clinical practice.