Background/aim: In this single-center study, we aimed to analyze texture features of primary uterine lesions on 18F-FDG PET/CT to predict lymph node metastases. Material and methods: Totally, 157 (mean age: 62 ± 10.2 years) patients were included in the analysis. Histopathological examination results were considered as the standard reference for nodal involvement. On $^{18}F-FDG PET/CT$ images, only the primary tumor was segmented. SUVmax, SUVmean, SUVpeak, MTV, and TLG of primary uterine lesions were calculated for analyses. For texture analysis first, second, and higher-order texture features were calculated. Results: Mean diameter of primary uterine lesions was calculated as 35± 18.1 mm. Lymph node metastases were detected in 19% of patients in histopathological examination of surgical materials. While 26 patients had pelvic lymph node metastases, 19 patients had additional paraaortic lymph node metastases. On radiomics analysis for 20 features, a significant difference was found between patients with and without lymph node metastasis. With using data mining methods GLZLM ZLNU, EntropyGLCM, Entropyhisto, GLRLM LRHGE, GLZLM HGZE, GLZLM SZHGE, GLRLM HGRE, GLRLM SRHGE were found significant radiomics features to predict lymph node metastasis with a diagnostic accuracy of 0.8. Conclusion: The radiomics analysis of intratumoral heterogeneity is a promising method for improving triage of the patients for lymph node dissection in endometrial carcinoma.