Skip to main content
. 2021 Dec 1;11:23202. doi: 10.1038/s41598-021-02495-6

Figure 5.

Figure 5

(Left) Multivariate CPH models of lacunarity, age at diagnosis, and abnormality radius for necrosis (n = 390), enhancement with necrosis (n = 402), and edema regions (n = 257). Lacunarity of both edema and enhancement with necrosis were significant predictors of overall survival in their respective models (p = 0.0007 and p = 0.042, respectively). Age at diagnosis, as expected, was consistently significant for survival. Radius was a significant predictor for regions of necrosis and enhancement with necrosis (p = 0.036). (Right) Corresponding Cox proportional hazard model of fractal dimension, age at diagnosis, and abnormality radii for necrosis, enhancement with necrosis, and edematous regions. Fractal dimension values of both edema and enhancement with necrosis were significant predictors of overall survival in their respective CPH models (p < 0.0001 and p = 0.0003, respectively). Age at diagnosis was consistently significant across all CPH models, while only enhancement with necrosis, and necrosis radii were significant (p = 0.0018 and p = 0.028, respectively). For a complete table including confidence intervals and significance values, see Supplement 4.