ESTRO 2020 Abstract Book

S854 ESTRO 2020

Conclusion In this work, we introduced a novel technique to assess the robustness of radiomics features against imaging artefacts. A set of robust features were identified and used to build a classification model for prostate cancer risk group. The positive results of this analysis provides enough evidence to support using PCa patient scans with and without fiducial markers for future multi-institutional validation studies. Reference [1] doi: 10.1016/j.ijrobp.2019.06.2504. PO-1575 Predicting doses to organs-at-risk in prostate cancer intensity modulated radiotherapy I. Mallick 1 , S. Saha 1 , M. Arunsingh 1 , A. Sarkar 2 , D. Guha Adhya 3 , R. Achari 1 1 Tata Medical Center, Department of radiation Oncology, Kolkata, India ; 2 Indian Institute of Technology, Department of Mechanical Engineering, Varanasi, India ; 3 Indian Institute of Technology, School of Medical Science and Technology, Kharagpur, India Purpose or Objective To predict doses to the rectum and bladder from the volumetric information available after segmentation in prostate cancer radiotherapy. Material and Methods We identified radiotherapy plan records of 98 patients treated with moderately hypofractionated radiotherapy with intensity modulated radiation therapy for prostate cancer to a prescribed dose of 60Gy in 20Fr. Dose volume histogram data was extracted from the planning system and text extraction tools were used to build a library of organ doses. We used volumes of the planning target volumes (PTVs) receiving 60 Gy and elective 44 Gy, the volumes of the rectum and bladder and the overlap volumes of the rectum and bladder with the PTVs to train a linear regression model and tested on a test data sub- set. The objective was to predict the relative volume receiving selected doses for each organ. Open source scientific computing libraries in Python were used for data extraction and calculations. Results Dose data could be extracted from the exported dose volume histogram files for all patients. Model evaluation parameters for the predicted relative volume receiving selected doses are shown in Table 1. The mean squared error (MSE) increased with lower dose-volumes, reflecting the overall variance at each dose level. Bladder doses- volumes could be accurately predicted with low median absolute errors (MAR), less than 2% at clinically important dose levels. Rectum dose-volume predictions were less accurate overall, mainly at the level of V59Gy which possibly reflects more dependence on planning constraints and priorities, than structure volumes.

PO-1576 Assessment of mpMRI-based radiomics tools in PCa for cancer aggressiveness prediction, AIRC IG- 13218 M. Pepa 1 , S.G. Gugliandolo 1 , L.J. Isaksson 1 , G. Marvaso 1 , S. Raimondi 2 , F. Botta 3 , S. Gandini 2 , D. Ciardo 1 , S. Volpe 1 , G. Riva 1 , D.P. Rojas 1 , D. Zerini 1 , P. Pricolo 4 , S. Alessi 4 , G. Petralia 4 , P. Summers 4 , A.F. Mistretta 5 , S. Luzzago 5 , F. Cattani 3 , O. De Cobelli 6 , E. Cassano 7 , M. Cremonesi 8 , M. Bellomi 4 , R. Orecchia 9 , B.A. Jereczek-Fossa 1 1 IEO European Institute of Oncology IRCCS, Division of Radiotherapy, Milan, Italy ; 2 IEO European Institute of Oncology IRCCS, Molecular and Pharmaco-Epidemiology Unit- Department of Experimental Oncology, Milan, Italy ; 3 IEO European Institute of Oncology IRCCS, Medical Physics Unit, Milan, Italy ; 4 IEO European Institute of Oncology IRCCS, Division of Radiology, Milan, Italy ; 5 IEO European Institute of Oncology IRCCS, Division of Urology, Milan, Italy ; 6 IEO European Institute of Oncology IRCCS, Department of Oncology and Hemato- Oncology, Milan, Italy ; 7 IEO European Institute of Oncology IRCCS, Breast Imaging Division, Milan, Italy ; 8 IEO European Institute of Oncology IRCCS, Radiation Research Unit, Milan, Italy ; 9 IEO European Institute of Oncology IRCCS, Scientific Directorate, Milan, Italy Purpose or Objective Radiomics involves testing the associations of a large number of quantitative imaging features with clinical characteristics. The purpose of this study was to extract a radiomic signature from axial T2-weighted (T2-W) multiparametric Magnetic Resonance (mpMRI) Imaging able to predict oncological and radiological scores in Give-me-five trial is a prospective phase II study designed for the treatment of PCa patients with ultra- hypofractionated radiotherapy scheduled in 5 fractions with 36.25 Gy delivered to the whole prostate and a concomitant boost of 37.5 Gy to the dominant intraprostatic lesion (DIL) identified by multiparametric MRI. T2-weighted (T2W) MRI sequences acquired with homogenous characteristics (0.59x0.59x3 mm 3 voxel size, XX Te, YY Tr) on a 1.5T Magnetom Avanto fit scanner (Siemens) were selected and the prostate gland contours were analyzed. The extraction of radiomic features (shape, first-order statistics and textural features) was performed using the IBEX software. We tested univariate and multivariate association of each radiomic features with T-stage (cT1 vs cT2), Gleason score (GS, 3+3 vs 3+4/4+3), extracapsular extension (ECE, 1/2 vs 3/4) score, Prostate Imaging – Reporting and Data System (PIRADS, 2/3 vs 4/5) score and risk class (intermediate vs low), and selected the feature with the lowest p-value in each cluster as representative. Statistical analysis was performed with SAS/STAT® software. Results Of the 65 prospectively enrolled patients, 49 T2W-MRI sequences fulfilled the inclusion criteria. Among the 1702 features extracted, 3 to 6 features with the highest predictive power were selected for each outcome. A logistic regression (machine learning) classifier was trained to predict clinical outcomes. Radiomic signature for prediction of high Gleason score included only GLCM3 texture features. Radiomic signature for prediction of cT2 stage as well as for 3/4 ECE score included first-order statistics intensity features. A GLCM3 texture feature was the most predictive feature for 4/5 PIRADAS score, with excellent predictive accuracy. Finally, radiomic signature for prediction of intermediate risk class included both GLCM3 texture and first-order statistics intensity features, with good predictive accuracy. Overall, the multivariable radiomic signature predicted oncological and radiological scores with AUC ranging from 0.74 to 0.94 (Table 1). prostate cancer (PCa). Material and Methods

Conclusion A simple linear regression model trained on a departmental plan library can provide useful prediction of dose-volumes in OARs from segmented volume information before planning. This can provide advance guidance to dosimetrists for setting specific plan constraints. Additional features from anatomy and treatment planning protocols will be tested to improve model performance.