Characterization of Colorectal Cancer Gross Target Volume with 18F-FDG PET/CT Image using the Second-Order Statistical Texture Analysis

Mohamed .O .M Ahmed, Sami Y. I. Awadain, Mohamed E. M. Gar-Elnabi

Abstract: The aim of this study is to characterize the colorectal region, including the rectum, tumor, and submucosal, using the Gray Level Co-occurrence Matrix (GLCM) and extract classification features from PET/CT with fluorine-18 fluorodeoxyglucose images. The GLCM technique was applied to identify variations in gray levels in the PET/CT images, which complements the features extracted from the images by estimating the distribution of sub-patterns using the Interactive Data Language (IDL) software. The results indicate that the combination of the Gray Level Co-occurrence Matrix and the extracted features achieved a classification accuracy of 100.0% for the rectum, 96.2% for the tumor, and 97.0% for the sub-mucosal, with an overall classification accuracy of 97.4 % for the colorectal area. These relationships are stored in a Texture Dictionary, which can be used in the future to automatically annotate new PET/CT images with the appropriate names for the colorectal regions.

Keywords: Colorectal cancer, 18F-FDG, PET/CT, Texture analysis