Feature Extraction for Pneumonia Diagnosis Using Conventional Methods

Medical image processing plays a crucial role in pneumonia diagnosis and classification using chest X-ray images. This research investigates key phases of medical image processing, focusing on Preprocessing, Feature Extraction, and Classification. Texture analysis methods, notably Local Binary Patterns (LBP) and Gray-Level Co-occurrence Matrix (GLCM), have been widely employed in medical imaging due to the inher-ent importance of textures in medical images. The significance of robust feature extraction methods lies in their ability to identify complex patterns indicative of various medical issues. By conducting experiments on a public chest X-ray dataset, this research employs established machine-learning classifiers to evaluate the efficacy of LBP and GLCM. The findings highlight the superior performance of GLCM in capturing nuanced textu-ral features, showcasing its potential as a principal contributor to enhancing pneumonia classification accuracy. However, the combination of both techniques produces even superior outcomes. LBP, known for its proficiency in capturing local texture patterns, complements GLCM, which excels in extracting global statistical textural information. The combination of these methods enhances the feature space, incorporating both local variations and long-range dependencies, thereby enriching the disease-discriminating patterns, particularly in pneumonia cases. From a wider perspective, texture analysis methods play a crucial role in enhancing the comprehension of medical images, thereby improving the accuracy and reliability of diagnoses.