ESTRO 2024 - Abstract Book

S602

Clinical - Breast

ESTRO 2024

Figure 2: ROC curve for SVM, sensitivity is the proportion of positives correctly identified, and specificity is the proportion of negatives correctly identified as such.

Table 1: The best prediction performance in Dosiomics model and DVH model

Group

Accuracy

Sensitivity

Specificity

AUC

Recall

F1score

DVH model

0.78

0.56

0.66

0.77

0.70

0.66

Dosiomics model

0.77

0.88

0.66

0.78

0.88

0.79

Conclusion:

Our preliminary results suggest that dosiomics features analysis yields valuable information for prediction of skin toxicity as one of the most important complication among breast cancer patients who received radiotherapy.(4-7)

Keywords: 3D-dose distribution, Machine learning

References:

1. Adachi T, Nakamura M, Shintani T, Mitsuyoshi T, Kakino R, Ogata T, et al. Multi ‐ institutional dose ‐ segmented dosiomic analysis for predicting radiation pneumonitis after lung stereotactic body radiation therapy. Medical Physics. 2021;48(4):1781-91.

2.

Ren W, Liang B, Sun C, Wu R, Men K, Xu Y, et al. Dosiomics-based prediction of radiation-induced

hypothyroidism in nasopharyngeal carcinoma patients. Physica Medica. 2021;89:219-25.