Program Information
Predicting Tumor Response in Esophageal Chemo-Radiation From Texture-Feature Analysis of FDG PET Images
P Galavis*, W Talcott , K Du , NYU Langone Medical Center, New York, NY
Presentations
SU-E-605-5 (Sunday, July 30, 2017) 1:00 PM - 1:55 PM Room: 605
Purpose: Despite advances in the multimodality care in esophageal cancers, in particular the combination of chemotherapy, radiotherapy and surgery, the 5-year overall survival rate remains only 15-34%. The ability to predict treatment response is therefore of great interest, with the potential to personalize cancer treatment. This project performs a comprehensive texture feature (TF) analysis from esophageal tumor [18F]FDG PET/CT images to establish their predictive value when compared with the PET Response Criteria in Solid Tumors (PERCIST).
Methods: Pre/Post chemo-radiation treatment [18F]FDG PET/CT images in sixty five patients were analyzed retrospectively. In all patients, the lesions were identified using nuclear medicine reports. The images were rigid-registered using CERR (a computational environment for clinical research), and segmented using thresholding method. Fifty features, based on the intensity histogram, second and high-order matrices, were extracted from the segmented regions from both image sets. The relative difference of SUVmax and of each TF were used as surrogates for treatment response. One-way ANOVA model of the intra-class correlation coefficient (ICC) and Spearman’s rank correlation coefficient (SC) were used to establish correlations between SUVmax and TF treatment response.
Results: Relative differences for fifty features were correlated with the corresponding SUVmax based on their ICC values, which were found in the range from 0.4 to 0.6. Two second-order and three high-order feature presented ICC = 0.6 (p<0.05). Spearman’s rank correlation coefficient ranged from -0.7 to 0.7. Two second-order and six high-order features presented 0.5≤SC≤0.6 (p<0.05)
Conclusion: Features with high ICC and SC values can be potentially used as additional metrics for treatment assessment. Hence, metabolic texture feature response provides a feasible approach for evaluating clinical outcomes in esophageal chemo-radiation.
Contact Email: