ESTRO 2023 - Abstract Book

S1867

Digital Posters

ESTRO 2023

Figure 2. The model parameters were: D50=19.33 Gy and gamma50=0.23.

PO-2084 Prediction of hormone status in breast cancer using 18F-FDG PET/CT radiomics features

B.Z. Devran 1 , O. Ayyıldız 2 , G. Özden 1 , H. Alkı ş 1 , T. Öne ş 3 , S. Özgüven 3

1 Marmara University, Radiation Oncology, Istanbul, Turkey; 2 Abdullah Gül University, Electrical and Electronics Engineering, Kayseri, Turkey; 3 Marmara University, Nuclear Medicine, Istanbul, Turkey Purpose or Objective This study aimed to predict the hormone status of patients with invasive ductal breast carcinoma by analyzing the 18F-FDG PET image’s maximal Standardized uptake value (SUVmax) and radiomics parameters. Materials and Methods Ninety-nine tumors in 97 breast cancer patients with pre-operative FDG PET images were included. Patients with non- ductal histology, excisional biopsy, multifocal tumors, T3/T4 tumors, and those who received pre-operative chemotherapy were not included. The median age was 51 (23-76). Hormone receptors and HER2 status were evaluated. Estrogen receptor (ER) was negative in 17 (%17) patients, negative progesterone receptor (PR) in 13 patients (positivity thresholds are used as %10 ¹ ), and negative human epidermal growth factor receptor 2 (HER2) status (<3+) in 78 patients. Radiomics features of primary breast lesions extracted by using Pyradiomics package and 3D-Slicer image computing software ² , ³ . The image processing software extracted 888 quantitative radiomics features from the segmented lesion area, including shape-based, first-order statistical, texture, and wavelet features. After the relief feature selection, train test settlers were created using 5-fold cross-validation, and a Linear SVM classifier was used to determine accuracy. 1. Yi M, Huo L, Koenig KB, Mittendorf EA, Meric-Bernstam F, Kuerer HM, et al. Which threshold for ER positivity? a retrospective study based on 9639 patients. Ann Oncol. 2014;25(5):1004-11. 2. Fedorov A, Beichel R, Kalpathy-Cramer J, Finet J, Fillion-Robin JC, Pujol S, et al. 3D Slicer as an image computing platform for the Quantitative Imaging Network. Magn Reson Imaging. 2012;30(9):1323-41. 3. van Griethuysen JJM, Fedorov A, Parmar C, Hosny A, Aucoin N, Narayan V, et al. Computational Radiomics System to Decode the Radiographic Phenotype. Cancer Res. 2017;77(21):e104-e7. Results The number of features has been reduced to 20 for ER and PR status using the Relief feature selection. Train test settlers were created using 5-fold cross-validation. Using a linear SVM classifier, the accuracy for ER and PR was 84% (AUC 0.83) and 75% (AUC 0.67), respectively. Five features were utilized for HER2 classification. The accuracy for 5-fold cross- validation using Linear Discriminant Analysis was 82% (AUC 0.75). Different radiomics features for ER, PR, and HER2 were significantly considerable. Conclusion Three-dimension imaging features from PET images were identified as potential biomarkers for distinguishing between ER, PR, and HER2 status of invasive ductal breast carcinomas. In the future, more extensive studies will be required to evaluate the findings.

PO-2085 A radiobiological model of the synergy between radiotherapy and immunotherapy

I. Gonzalez-Crespo 1,3 , A. Buñuel-Muriscot 1 , A. Gomez-Caamaño 2 , O. Lopez Pouso 3,1 , J.D. Fenwick 4 , J. Pardo-Montero 1,5

1 Health Research Institute of Santiago, Group of Medical Physics and Biomathematics, Santiago de Compostela, Spain; 2 Clinical University Hospital of Santiago, Department of Radiation Oncology, Santiago de Compostela, Spain; 3 University of Santiago de Compostela, Department of Applied Mathematics, Santiago de Compostela, Spain; 4 Institute of Translational