ESTRO 2024 - Abstract Book

S5138

Physics - Radiomics, functional and biological imaging and outcome prediction

ESTRO 2024

better identify the ORN patients for AXB dose distribution. Due to limited number of patients enrolled for this explorative study, only training dataset was included in the analysis.

Dose-calculation algorithm

Accuracy [%]

Specificity [%]

Sensibility [%]

Precision [%]

AAA

22

33

11

14

AXB Dw

83

100

67

100

AXB Dm

83

67

100

75

Table 1 – ORN prediction model performance evaluated in terms of specificity, sensibility, precision and accuracy.

Conclusion:

It is already well-known that there are important differences between AAA and AXB in VMAT planning for H&N [2]. Even if based with a small patients’ cohort and without an evaluation and test dataset, dosiomic features could further improve the granularity understanding of dose distribution additionally to the classical DVH level. Indeed, it looks to select specific dosiomic features able to identify the optimal prediction model to foresee ORN, highlighting differences between AAA and AXB dose distribution. Further analysis is ongoing, increasing the number of the training data set, including an evaluation and test dataset, and also considering multiclassing parameter classification based on the ORN CTCAE scoring at peak stage.

Keywords: dosiomics, H&N toxicity, Acuros XB and AAA

References:

[1] Placidi L et al. On dose cube pixel spacing pre-processing for features extraction stability in dosiomic studies. Phys Med. 2021 Oct;90:108-114. doi: 10.1016/j.ejmp.2021.09.010. Epub 2021 Sep 30. PMID: 34600351.

[2] Muñoz-Montplet C et al. Dosimetric impact of Acuros XB dose-to-water and dose-to-medium reporting modes on VMAT planning for head and neck cancer. Phys Med. 2018 Nov;55:107-115. doi: 10.1016/j.ejmp.2018.10.024. Epub 2018 Nov 15. PMID: 30471814.