Classification of Normal Brain Tissues in Computed Tomography Radiographs Using Image Texture Analysis

Abstract

This study was conducted to classify normal brain tissue in computed tomography (CT) radiographs using image texture analysis; into white matter, grey matter CSF and bone using texture feature extracted from CT images. The extracted feature classified using linear discriminate analysis. The data obtained from Modern Medical Center [Sudan/ Khartoum], and King Khalid Hospital [KSA/ Kharj], the study samples were consisted of 140 patients’ images with normal brain underwent CT brain examination. The data was collected in period from July 2017 October 2017. The images were analyzed using Interactive Data Language IDL software. The results of this study showed that the overall accuracy of classification was 97.0%; with the accuracy of classification for Bone was 99.5%, White matter was 97.3%, Gray matter was 95.3%, and CSF was 96.2%. Study showed texture analysis using first order statistics can be used to classify normal brain tissues with high degree of accuracy; therefore abnormal tissues can be identified with an acceptable accuracy.