Three-dimensional PET Image Texture Analysis Of Esophageal Squamous Cell Carcinoma

ObjectiveTo explore the relationship of new PET image parameter,18F-DG uptakeheterogeneity assessed by texture analysis, with common metabolic parametermaximum standardized uptake value (SUVmax) and to investigate their correlationswith the American Joint Committee on Cancer (AJCC) staging system.MethodsForty consecutive patients with esophageal squamous cell carcinoma (ESCC)were enrolled. All the patients received whole body preoperative18F-FDG PET/CTscanning. Intra-tumor18F-FDG uptake heterogeneity was assessed bythree-dimensional image texture features (entropy and energy) using MATLABsoftware. The texture parameter_s were correlated with SUVmax, tumor-nodemetastases (TNM) classification, histologic grade, tumor location and AJCC stage.ResultsTumor_s with higher SUVmaxtend to be more heterogeneous on18F-FDG uptake.There were significant correlations between T stage and SUVmax(r_s=0.390P=0.013),entropy (r_s=0.693P<0.001) and energy (r_s=-0.469P=0.002); N stage and SUVmax(r_s=0.326P=0.04), entropy (r_s=0.501P=0.001) and energy (r_s=-0.413P=0.008).AJCC stage also correlated significantly with all metabolic parameter_s. ROC curve demonstrated an entropy of4.699to be the optimal cut-off points for detecting tumor_sabove stage IIb with an AUC of0.789(P<0.001). There was no significant correlationbetween any metabolic parameter and histologic grade or tumor location.ConclusionsThis study provides initial evidence for a relationship between new tumor uptakeheterogeneity parameter and commonly used simplistic SUV parameter and tumorstage. Our findings may suggest a complementary role of these parameter_s to theexisting role of PET in staging and prognostication for ESCC.