ESTRO 2021 Abstract Book

S1526

ESTRO 2021

features together, scaling and normalization were applied. An unsupervised analysis based on hierarchical clustering was used together with the Fisher’s exact test to estimate statistical significance. Results We trained and evaluated the machine learning performance within a 5-fold cross validation scheme. At each iteration, a different test set consisting of an equivalent number of subjects with and without toxicity was used. Individual dose, T1w, T2w and FLAIR data were not predictive, leading to average accuracies of 0.44±0.21, 0.56±0.16, 0.59±0.16 and 0.41±0.07, respectively. While combining all multiparametric information a better prediction performance was obtained: 0.62±0.09.

Additionally, results were obtained with statistical features analysis. In fig.1 is reported the correlation matrix used to evaluate the degree of redundancy of feature data in the dataset. The result of Fisher’s test is displayed by Kaplan-Meier plot in fig.2. All the features available were predictive with a specificity of 0.79 and sensitivity of 0.65.

Conclusion Promising results have been obtained applying Random Forest classifier and statistical analysis to radiomic and dosiomic features of paediatric patients affected by. This multimodal and multiparametric approach could have a large impact for precision medicine, as radiomic biomarkers are non-invasive and can be applied to imaging data that are already acquired in clinical settings. PO- 1800 Exploring hypoxia in prostate cancer with T2-weighted MRI radiomics and pimonidazole scoring. Michelle Leech 1 , Ralph Leijenaar 2 , Tord Hompland 3 , John Gaffney 4 , Heidi Lyng 5 , Laure Marignol 1 1 Trinity St. James's Cancer Institute, Applied Radiation Therapy Trinity, Trinity College, Radiation Therapy, Dublin, Ireland; 2 Oncoradiomics SA, Oncoradiomics, Liége, Belgium; 3 Radiumhospitalet Oslo University Hospital,, Radiation Biology, Oslo, Norway; 4 St. Luke's Radiation Oncology Network, Radiation Oncology, Dublin, Ireland; 5 Radiumhospitalet Oslo University Hospita, Radiation Biology, Oslo, Norway Purpose or Objective High levels of hypoxia are associated with a poorer prognosis in prostate cancer. Regions of a tumour with a high oxygen concentration are believed to be up to three times more amenable to radiation therapy than are hypoxic regions. Imaging techniques such as DCE MRI, BOLD- MRI or PET using hypoxia-seeking ligands such as FMISO, FAZA and 18 F]HX4 are not able to fully sample the tumour microenvironment and both acute and chronic hypoxia are believed to be micro-regional. Radiomics, which involves advanced image analysis and high throughput extraction of mineable precise quantitative imaging descriptors or features that serve as non- invasive prognostic or predictive biomarkers may be an alternate to overcome this issue in the identification of hypoxia in prostate cancer. Materials and Methods The prostatic index lesion of 88 intermediate or high-risk prostate cancer patients’ T2-weighted MRIs were analysed in this study. All patients received the hypoxia marker pimonidazole (PIMO) prior to radical prostatectomy. MRI scans were acquired adhering to ESUR 2012 guidelines. PIMO hypoxic scores were assigned by an experienced pathologist who was blinded to MRI. Radiomics feature extraction was performed using an evaluation version of RadiomiX (RadiomiX Research Toolbox version 20180831 (OncoRadiomics SA, Liège,