ESTRO 2021 Abstract Book

S1412

ESTRO 2021

factors including age, gender, WHO grade, Isocitrate dehydrogenases (IDH) mutation status and treatment after surgery were collected. We use the patient model and region model for recurrence prediction. In the patient model, we use the whole tumor region for radiomics calculation. For the region model, we analysis the 40 recurrence patients’ recurrence and non-recurrence regions. The least absolute shrinkage and selection operator (LASSO) regression model was conducted for data dimension reduction, feature selection, and radiomics feature analysis. Internal validation was assessed. Furthermore, multivariable logistic regression analysis was used to develop the predicting model by combining the radiomics signature and independent clinical factors Results In total, 129 patients were included, among which 40 patients had recurrence. The median follow-up time was 27.4 (range, 2.6–79.2) months. For clinical factor, the gender and concurrent chemotherapy have no significant influence. There are significant differences in age (p=0.037), WHO grade (p=0.001), IDH status (p=0.03), radiotherapy-interruption (p=0.017) and adjuvant chemotherapy (p=0.009) between recurrence and non-recurrence patients. In the patient model, the radiomics signature was associated with the event of recurrence (P < 0.001 for both training and validation cohorts, respectively). The AUC values for each radiomics signature of three regular MR image sequence (T1WI, T2WI, contrast-enhanced T1WI ) were showed in Table 1. Good calibration was observed for the prediction probability for recurrence patients (Figure 1). The calibration curve for the probability of local recurrence region prediction also demonstrated good agreement between prediction and observation in the prediction model.

Table 1. Prediction power analysis in patient model

Figure 1. Calibration curve for T1WI, T2WI, contrast-enhance T1WI of patient model

Conclusion We identified radiomics feature derived from brain MRI that presented potential in predicting recurrence in glioma patients. This preliminary finding allows possibility in exploring risk prediction models for early identification of recurrence for such patients. Further investigation is necessary to include expanded sample size investigation and external multicenter validation. PO-1686 A novel semi auto-segmentation method for head and neck adaptive radiotherapy Y. Gan 1 , J.A. Langendijk 2 , E. Oldehinkel 1 , D. Scandurra 1 , N. Sijtsema 1 , Z. Lin 3 , S. Both 1 , C. Brouwer 1 1 University of Groningen, University Medical Center Groningen, Radiation Oncology, Groningen, The Netherlands; 2 University of Groningen, University Medical Center Groningen, Radiation Oncology, Groningen, The Netherlands; 3 Shantou University,Cancer Hospital of Shantou University Medical College, Radiation oncology, Shantou, China Purpose or Objective Robust segmentation of organs at risk (OARs) plays a crucial role in radiotherapy. Segmentation becomes increasingly tedious and time-consuming with the introduction of adaptive radiotherapy (ART) which requires repeat CT (rCT) scan and OAR re-segmentation. The purpose of this work was to assess the accuracy of different methods for re-segmentation of different OARs, considering the variability in mean dose (D mean ) and Normal Tissue Complication Probability (NTCP). Based on these results, a semi auto-segmentation(SAS) method