Fig. 5: Construction and validation of a radiomic signature for predicting RE-related subtypes in bladder cancer. | npj Digital Medicine

Fig. 5: Construction and validation of a radiomic signature for predicting RE-related subtypes in bladder cancer.

From: Integrative molecular–radiopathomic characterization of AIRE-driven immune subtypes in bladder cancer

Fig. 5

A The top 40 radiomic features were selected using the mRMR algorithm from a total of 3562 features extracted from T2WI and axial DCE sequences. B LASSO regression identified 9 features with non-zero coefficients for signature construction. C The LASSO model demonstrated favorable predictive performance with an AUC of 0.79 in the training set and 0.75 in the validation set. D The radiomic model showed high sensitivity and specificity in distinguishing RE-related subtypes. E, F Confusion matrices confirmed accurate classification in both training and validation cohorts. G, H Tumor morphological differences between subtypes were observed: high RE score tumors exhibited exophytic, papillary-like patterns, while low RE score tumors displayed endophytic, flat-like patterns.

Back to article page