Development and Evaluation of an Open-Source Software Package “CGITA” for Quantifying Tumor Heterogeneity with Molecular Images

Figure 3

Parametric images of textural features computed from a single patient compared to the original PET image. The PET image, shown in the top row, is displayed between SUVs of zero and twenty. Heterogeneity indices 1 to 4 represent, respectively, the contrast, dissimilarity, entropy, and inverse difference moment calculated from the cooccurrence matrix. Note that the parametric images appear different according to the spatial variation in voxel intensity. Furthermore, different index images display different tumor heterogeneity patterns.