ESTRO 2024 - Abstract Book

S4580

Physics - Machine learning models and clinical applications

ESTRO 2024

Conclusion:

In conclusion, this study underscores the effectiveness of the combined approach incorporating Surface Mapping scores and machine learning algorithms to predict resimulation decisions in prostate cancer radiotherapy. The results emphasize the significant potential for improving the accuracy of resimulation probability predictions and facilitating timely resimulation decisions in the clinical setting. SM-based ML models could play a crucial role in optimizing resimulation decisions by identifying relevant factors currently used in clinical practice. In future work, there is room for refining the model further, with the aim of expediting informed earlier resimulation decisions and thereby reducing the risk associated with undesirable treatment set-ups, ultimately contributing to enhanced patient care.

Keywords: Surface mapping, Machine learning, Resimulation

References:

1. M Kassel, CY Shang, G Evans, TR Williams: A Novel Analytic Method of Automatically Assessing Beamline Geometric Range Variations for CBCT-Guided Intensity-Modulated Proton Therapy (IMPT) for Localized Prostate Cancer. Int J Rad Onc Bio Phy 2023; 117 (2), e715

2. Bishop, C. M. (1995). Neural Networks for Pattern Recognition. Oxford University Press.

3. Breiman, L. (2001). Random Forests. Machine Learning, 45(1), 5-32.

4. Friedman, J. H. (2001). Greedy Function Approximation: A Gradient Boosting Machine. The Annals of Statistics, 29(5), 1189-1232.