CXR Images and ELM based Advanced Method for Classification of Multivariant Pneumonia

Abstract

In the current era of COVID19, pneumonia diagnosis and treatment are of utmost importance. Although analysis of chest x-ray image a crucial role in pneumonia diagnosis, even experienced radiologists face difficultiesin accurately diagnosing pneumonia because of CXR images and human judgment. An automated system that employs deep learning (DL) technology could significantly enhance the reliability of CXR image analysis. While several existing methods have been proposed to find the pneumonia but found difficulty with accuracy issues. Here it is t proposed an automated pneumonia detection system utilizing images from. Extreme Learning Machine (ELM)). The researcher used ELM classification, ELM combined with hybrid CNN-PCA feature extraction, and his CNN-PCA-ELM using his CXR images enhanced by contrast-limited adaptive histogram equalization (CLAHE). Here, two different models are tested. The final model produced exceptional results, achieving 98.32% accuracy and 98% recall for multiclass pneumonia classification, and 100% recall and 99.83% accuracy for binary classification. The proposed method outperformed existing methods in terms of accuracy, precision, recall, and other relevant benchmarks. These findings highlight the potential of the proposed system as a robust tool for diagnosing pneumonia. It is expected that the implementation of this automated pneumonia detection system in clinical practice will improve the accuracy of pneumonia diagnosis and ultimately lead to better patient outcomes.