ESTRO 2025 - Abstract Book

S3404

Physics - Machine learning models and clinical applications

ESTRO 2025

Figure 2a presents the calculated risks of microscopic metastases per LNL, given a certain observation of macroscopic metastases in upstream LNLs on PET-CT, in early T-category. Results show that patients with no visible LNL involvement on PET-CT still carry a 13% risk for microscopic metastases in LNL1. Figure 2b illustrates the evolution of risk of microscopic metastases in LNL2 and LNL3 with primary T-category. For late T-category patients with upstream LNLs involvement on PET-CT, risk of LNL2 and LNL3 involvement increased from 8% and 13% to 16% and 29%, respectively.

Conclusion: A probabilistic model to assess the risk of microscopic metastases in axillary LNL 1 to 3 in eBC patients was constructed and its use was illustrated in combination with PET-CT for nodal staging. Ongoing work includes expansion of the dataset and model to supraclavicular, internal mammary, and interpectoral nodes, with the inclusion of other staging modalities such as sentinel lymph node biopsy and patient and disease-specific factors such as tumor subtype.

Keywords: Breast, CTV, radiotherapy

References: [1] Ludwig, R., Pouymayou, B., Balermpas, P. et al. A hidden Markov model for lymphatic tumor progression in the head and neck. Sci Rep 11, 12261 (2021). doi:10.1038/s41598-021-91544-1 [2] Rosen, P P et al. “Discontinuous or "skip" metastases in breast carcinoma. Analysis of 1228 axillary dissections.” Annals of surgery vol. 197,3 (1983): 276-83. doi:10.1097/00000658-198303000-00006 [3] Liang, X et al. “MRI and FDG -PET/CT based assessment of axillary lymph node metastasis in early breast cancer: a meta- analysis.” Clinical radiology vol. 72,4 (2017): 295-301. doi:10.1016/j.crad.2016.12.001