ESTRO 2025 - Abstract Book

S3845

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2025

Figure 1: The Confusion Matrix of the best results obtained with Random Forest, using automatically extracted augmented data.

Figure 2: The Receiver Operating Characteristic Curve of the best results obtained with Random Forest, using automatically extracted augmented data.

Conclusion: This project highlights the transformative potential of predictive models in medicine, enabling outcome forecasting through advanced health data analysis. The best results were achieved with balanced data and automated feature selection, underscoring the importance of addressing techniques to handle imbalanced data, and using efficient methods for feature selection. Furthermore, it emphasizes the significance of combining radiomics with clinical and