356 chapter 5 Table 6. Radiomic analysis: dimensionality reduction and predictive performance of radiomic models for all [18F]FDG-positive nodules, non-Hürthle cell nodules and Hürthle cell nodules All [18F]FDG-positive nodules (n=84) Non-Hürthle cell nodules (n=56) Hürthle cell nodules (n=28) [18F]FDG-positive nodules > 64 voxels (n=66) PET/CT PET PET/CT PET PET/CT PET PET KMO, median (range)a 0.967 (0.965-0.972) 0.938 (0.927-0.946) 0.977 (0.975-0.980) 0.964 (0.960-0.970) 0.978 (0.973-0.981) 0.952 (0.943-0.961) 0.946 (0.934-0.954) Cumulative variance all factors, median (range)a 0.654 (0.637-0.680) 0.796 (0.774-0.821) 0.518 (0.482-0.560) 0.673 (0.618-0.712) 0.306 (0.273-0.342) 0.515 (0.466-0.563) 0.691 (0.677-0.720) Retained factorsb 6 factors: - Nodule size - Variance of the intensity histogram on CT - High intensity on PET - Fine textures on CT - Mutual information between different regions on CT - Large areas on PET 6 factors: - Entropy of the intensity histogram - Nodule size - High intensity on PET - Variance in area size - Total lesion glycolysis - Small areas with low grey levels 4 factors: - High intensity on PET - Nodule size - Variance of the intensity histogram on CT - Variance in voxel dependency on CT 4 factors: - High intensity on PET - Entropy of the intensity histogram - Nodule size - Large areas 2 factors: - Variance of the intensity histogram on CT - Entropy of the intensity histogram on PET 2 factors: - Nodule size - Entropy of the intensity histogram 5 factors: - High intensity on PET - Nodule size - Entropy of the intensity histogram - Variance in area size - Total lesion glycolysis Test AUC (95% CI) 0.461 (0.289-0.633) 0.445 (0.290-0.600) 0.466 (0.236-0.695) 0.519 (0.298-0.740) 0.537 (0.208-0.867) 0.694 (0.461-0.926) 0.421 (0.303-0.538) KMO, Kaiser-Meier Olkin measure, AUC, area under the receiver operating characteristic curve. a: The KMO and the cumulative variance of all factors are expressed as median and range of all 20 folds. b: One factor was retained per ten subjects in the training set.
RkJQdWJsaXNoZXIy MTk4NDMw