ESTRO 2024 - Abstract Book

S4527

Physics - Machine learning models and clinical applications

ESTRO 2024

Conclusion:

A combined end-to-end framework was utilised in this study to classify RT structures. Our experiments demonstrate that the use of multi-modal data can improve the performance compared to univariate models. The developed model has the potential to be utilised to retrospectively classify inconsistent structure names into standardised nomenclature, adhering to national and international consensus guidelines. Such automated systems will have potential to address inconsistencies among RT structure names in a platform that utilises multiple large-sized clinical practice databases and for real-time quality assurance use, reducing the risk of mislabelled data. We therefore propose the development of this work using breast cancer data as a pilot that could be successfully extended for other tumour sites in the future.

Keywords: Radiotherapy Dataset, Nomenclature Standardisation

References:

[1] Schuler T, Kipritidis J, Eade T, Hruby G, Kneebone A, Perez M, Grimberg K, Richardson K, Evill S, Evans B, and Gallego B. Big data readiness in radiation oncology: An efficient approach for relabeling radiation therapy structures with their TG-263 standard name in real-world data sets. Advances in Radiation Oncology, 4(1), 2018.

[2] Mayo CS et al. American association of physicists in medicine task group 263: Standardizing nomenclatures in radiation oncology. International Journal of Radiation Oncology, Biology, Physics, 100(4):1057–1066, 2018.

[3] Jarrett D, Stride E, Vallis K, and Gooding MJ. Applications and limitations of machine learning in radiation oncology. British Journal of Radiology, 92(1100), 2019.

[4] Lustberg T, van Soest J, Gooding M, Peressutti D, Aljabar P, van der Stoep J, van Elmpt W, and Dekker A. Clinical evaluation of atlas and deep learning based automatic contouring for lung cancer. Radiotherapy and Oncology, 126(2):312–317, 2018.