TY - JOUR T1 - Entropy characterizes intratumoral hypometabolism in FDG PET JF - Journal of Nuclear Medicine JO - J Nucl Med SP - 1779 LP - 1779 VL - 56 IS - supplement 3 AU - Fanny Orlhac AU - Michael Soussan AU - Soraya Djelbani AU - Jennifer Tordjmann AU - Irene Buvat Y1 - 2015/05/01 UR - http://jnm.snmjournals.org/content/56/supplement_3/1779.abstract N2 - 1779 Objectives Recent studies suggest that texture indices (TI) can assess tumor heterogeneity and predict treatment response and/or patient survival. Yet, the actual physiological meaning of these TI is unknown. We investigated whether TI could reflect the presence of hypometabolic areas (hMA) in FDG-avid tumors as visually assessed by observers.Methods Forty-nine treatment-naïve patients with breast cancer underwent a PET/CT scan at 77±9 min after injection of 18F-FDG (3 MBq/kg). The primary lesion was visually assessed independently by three experienced nuclear physicians separating lesions into 2 groups: including focal hMA or not. A second consensus reading involving all 3 observers led to the final reference classification. Each lesion was also segmented by a semi-automatic method. Metabolic Volume (MV), maximum Standardized Uptake Value (SUVmax) and 6 TI were computed. The ability of each parameter to distinguish between lesions with or w/o hMA as defined by the consensus was assessed using Wilcoxon’s tests and ROC analyses.Results The interobserver agreement was moderate (к between 0.269 to 0.492) calling for the need to identify an objective mean for detecting hMA in tumors. The consensus reading yielded 25 lesions with hMA against 24 lesions w/o. Three TI, SUVmax and MV were related with hMA as visually assessed by the consensus (p≤0.003). hMA lesions had lower Low Gray-level Zone Emphasis, higher Entropy and higher High Gray-level Zone Emphasis than non-hMA lesions. Entropy led to an Area Under the ROC Curve (AUC) of 0.877, which was higher than AUC of any other index, including SUVmax (AUC=0.772) and MV (AUC=0.820).Conclusions Identification of lesions with hMA is strongly observer-dependent. Three TI could identify hMA lesions, among which Entropy was the most correlated with the presence of hMA. As hMA has been shown to be a prognostic factor in oncology, Entropy might be considered as a hMA biomarker. ER -