In this contribution, we study the dependence of radiomics features extracted from 18F-fluorodeoxyglucose -positron emission tomography/computed tomography (18F-FDG PET/CT) on different SUV discretization intensity values in patients with liver cancer type hepatocellular carcinoma (HCC). Materials and methods. 35 radiomic features were extracted from PET images using semiautomatic method through 4 discretization intensity values with number of grey levels = (32, 64, 128, 256). Determination of bin within dour different. One way analysis of variance (ANOVA) was used to obtain whether there is significant difference between radiomics features at the four discretization values.Results. By comparing the average and variance of radiomics through the four discretization values, only the first order skewness and kurtosis are stable through the different discretization values, where all the other features demonstrated significant difference between features in the 4 values. Spearman correlation coefficient between radiomics features and tumors parameters calculated.Conclusion. The discretization level of 256 have been found to be the optimum discretization value to extract the radiomics features.