ESTRO 2025 - Abstract Book

S4009

Radiobiology - Tumour radiobiology

ESTRO 2025

application of fuzzy region and FIS did not significantly outperform traditional methods (0.659±0.026 and 0.661±0.030 vs. 0.668±0.030, with p -values of 0.029 and 0.118), their combined use yielded a higher C-index and reduced error range (0.677±0.027, p =0.019) (illustrated in Figure 2B).

Conclusion: The fuzzy statistical approach can reduce the impact of contour uncertainty by fuzzifying the tumor ROI boundaries and enhance the ability to characterize the heterogeneity features of the tumor boundaries and microenvironment. It has demonstrated the capability to provide a more comprehensive image phenotype of the tumor and achieves better prognosis prediction performance for the radiomics analysis of oropharynx cancer.

Keywords: Radiomics, Oropharyngeal cancer

References: [1] Teng, Xinzhi, et al. "Building reliable radiomic models using image perturbation." Scientific Reports 12.1 (2022): 10035. [2] Grahovac, Marko, et al. "Machine learning predictive performance evaluation of conventional and fuzzy radiomics in clinical cancer imaging cohorts." European Journal of Nuclear Medicine and Molecular Imaging 50.6 (2023): 1607-1620. [3] Cao, Jin, et al. "Fuzzy inference system with interpretable fuzzy rules: Advancing explainable artificial intelligence for disease diagnosis—A comprehensive review." Information Sciences 662 (2024): 120212.