RESP dataset construction with multiclass classification in respiratory disease infection detection using machine learning approach

The respiratory illness issue has put significant and growing strain on the world's medical systems. Medical image processing can help with illness diagnosis, treatment, and early detection. Deep Learning (DL) models would assist physicians and radiologists in providing prompt diagnostic support to patients. Our study focuses on case detection models for respiratory disorders such as COVID-19, pneumonia, and TB that leverage deep learning. The goal of the proposed concept is to understand the importance of data balancing, data augmentation, and segmentation in the clinical field, to improve image data balancing using data augmentation and edge detection techniques, to improve radiology image preprocessing to locate regions of interest (ROI), and to construct custom-built Deep Neural Networks (DNN) RESP_DNN in diagnosing respiratory illnesses using Machine Learning approaches. The proposed approach creates RESP datasets with enhanced radiological image preprocessing to find ROI, and then modifies the layered Deep neural network architecture for deep learning. The suggested approach correctly diagnoses five classes: COVID-19, pneumonia, lung opacity, tuberculosis, and normal, with an accuracy of 95.52%. When compared to recently published strategies in the literature, the suggested methodology outperforms them in terms of accuracy. The experimental goal of our work is to help in the categorization and early detection of respiratory diseases. This high degree of accuracy is a unique and potentially significant resource that enables radiologists to rapidly detect and diagnose respiratory disorders like COVID-19, Pneumonia, Tuberculosis, Lung Opacity patients using a deep learning model.