Fig. 2: Response prediction performance and explainability. | BJC Reports

Fig. 2: Response prediction performance and explainability.

From: Radiomics and outcome prediction to antiangiogenic treatment in advanced gastroenteropancreatic neuroendocrine tumours: findings from the phase II TALENT trial

Fig. 2: Response prediction performance and explainability.

Performance of the radiomics and radiomics-clinical models to predict response to treatment using: a Receiver operating characteristic (ROC) curves for median progression-free survival (PFS) prediction and b Kaplan–Meier curves for PFS probability. Explainability of the radiomics signature: c Associations between vascular factor expression and radiomics score. The cluster A corresponds to highly vascularises tumours with high-radiomics score, while cluster B includes tumours with poor vascularisation and low-radiomics score. d Visualisation of tumours with high-radiomics score (i.e., more spherical, heterogenous and highly enhancing) and long PFS vs. low-radiomics score and short PFS. Spearman’s rank correlation analysis (p < 0.01).

Back to article page