ESTRO 37 Abstract book

S1159

ESTRO 37

Results Median tumor volume was 1cm 3 . Table 1 lists the most relevant correlations of clinical parameters, textural features and MRI-parameters.

over the radiotherapy course. The change of CTN was linearly correlated with radiation doses (mean R2 = 0.889 ± 0.164), while the correlation between CTN changes and TV reduction ratio was weaker (R2 = 0.257). For patients with response, the TV and CTN reduced by 32.01% (8.46%- 61.67%) and 28.44±13.12 Hu, which were significantly higher than those in the non-response patients with 23.20% (-15.57%-38%) and 19.63±8.67 Hu (P = 0.026 and P = 0.005), respectively. ROC curve illustrated that both TV shrinkage ratio (AUC = 0.693, P = 0.016) and CTN variation (AUC = 0.666, P = 0.037) could predict treatment response. The area under curve for combination of TV and CTN was larger than any one alone (AUC = 0.751, P = 0.002). TV reduction during the period of D6-D11 (AUC = 0.667, P = 0.036) and CTN changes between D11 and D16 (AUC = 0.753, P = 0.002) demonstrated significance difference between response and non-response. Conclusion For NSCLC tumor target, CTN variation was linearly correlated with the radiation dose received. The changes of TV and CTN obtained from CBCT images have the potential capability to predict early response of NSCLC, which could be able to identify NSCLC patients who benefit from CRT. The prediction capability may be improved by the combination of the changes on TV and CTN. EP-2104 Influence of androgen deprivation therapy on MRI parameters and textural features in prostate cancer M. Daniel 1,2 , P. Kuess 1,2 , P. Andrzejewski 2,3 , P. Baltzer 2,4 , S. Polanec 2,4 , F. Dragschitz 1,2 , G. Goldner 1,2 , D. Georg 1,2 1 Medical University of Vienna, Department of Radiotherapy, Vienna, Austria 2 Medical University of Vienna, Christian Doppler Laboratory for Medical Radiation Research for Radiation Oncology, Vienna, Austria 3 EBG MedAustron GmbH, Medical Physics, Wiener Neustadt, Austria 4 Medical University of Vienna, Department of Biomedical Imaging and Image-Guided Therapy, Vienna, Austria Purpose or Objective The fraction of patients receiving androgen deprivation therapy (ADT) is increasing. It is known that ADT has an impact on T2w signal intensity and increases the apparent diffusion coefficient (ADC) in the lesion. These factors challenge the use of MRI techniques for lesion characterization and response assessment in prostate cancer. We investigate the utility of Haralick textural features (TF) in this context. Material and Methods 44 multiparametric MRI datasets of primary prostate cancer patients were collected for a retrospective study. All patients received external beam radiotherapy between 11/2013-01/2017 and were scanned on a 3T MRI prior radiotherapy. 29/44 patients received ADT before MRI. The imaging protocol included T2w images, a DCE and a DWI sequence. 10 of these datasets (5 with ADT/5 without ADT) were selected for this current pilot analysis. The tumor lesions were delineated by a radiologist specialized in prostate cancer. Tumor volume was assessed and the subsequent analysis was restricted to this area. Analysis was performed utilizing the MICE Toolkit (www.gentleradiotherapy.se, Umea, Sweden). Mean, max, min and histogram parameters were assessed as well as 19 different Haralick TF from T2w and ADC. Correlation coefficients of TF, MRI-parameters and clinical data (PSA initial , PSA nadir , Gleason score) were calculated. After splitting the data into the two patient groups, a comparison was performed to investigate the influence of ADT.

Gleason score, PSA initial showed non-significant weak positive correlations with each other. Correlations of textural features with tumor volume were observed. However, this might be caused by a size dependency of the features in question. T2w parameters did not correlate with T2w TF, while for ADC max strong negative correlations were found with ADC Contrast, DifferenceVariance and DifferenceEntropy . The DCE- derived parameters showed significant correlations with TF as well. Table 2 summarizes parameters of interest for the split patient groups. and PSA nadir

ADC parameters showed the expected higher values in patients receiving ADT. However, the difference did not reach statistical significance. The results for four Haralick TF showed significant differences between the two patient groups. Conclusion The findings of this pilot analysis indicate that the TF T2w SumAverage , Autocorrelation and ClusterShade as well as the ADC DifferenceEntropy are influenced by ADT. These preliminary results motivate further investigation of this behavior and analysis of the remaining datasets. Furthermore, they indicate a possible relevance of the residual TF in characterizing tumors and predicting treatment response for prostate cancer patients undergoing radiotherapy as well as ADT.