Development of GLCM Method in Calculate Entropy Value for Digital Visualization in Identifying Childhood Pneumonia Based on Chest X-Ray Images

Abstract

Pneumonia can affect people of all ages, especially children. One way to identify pneumonia is by using medical equipment through radiological examinations such as chest X-rays. This study proposes the development of an entropy formula found in the gray level co-occurrence matrix (GLCM) texture extraction method to automatically detect pediatric chest X-ray results in identifying pneumonia. The pre-processing stage is tested with several steps, including converting RGB to grayscale, adaptive histogram equalization (AHE), filtering, Otsu thresholding, image inversion, and automatic image cropping. After preprocessing is the segmentation stage that conducted by processing the image from the cropping results. The testing process in the segmentation stage includes contrast enhancement, Otsu multi-thresholding, border clearing, and image segmentation. The results from the segmentation process are then followed by the extraction stage. The extraction stage focuses on developing the entropy value found in GLCM, referred to as the entropy value algorithm with gray level co-occurrence matrix (EVAGLCM). The key contributions of this study lie in the advancement of digital image processing methods for the accurate identification of childhood pneumonia through improved texture feature extraction. This study compares the developed entropy value with several previous studies. The development of this entropy value is then followed by the classification stage using a support vector machine (SVM). The accuracy achieved in this study was 97.5%, meaning it was able to accurately detect 390 images out of 400 images. This indicates that the entropy value calculation using the EVA-GLCM formula and classification using SVM can provide more accurate output with a higher accuracy rate.