ESTRO 2023 - Abstract Book

S1872

Digital Posters

ESTRO 2023

Simulated annealing resulted in models with comparable predictive accuracy to out-of-box optimization techniques (Cystitis: SA=57.3±6.6%, TAN=53.7±16.2%, GHC=55.4±6.4%, CLO=56.4±6.5%. Proctitis: SA=55.2±7.0%, TAN=42.4±14.5%, GHC=53.8±5.1%, CLO=53.3±5.8%). The benefit of the simulated annealing technique is the highly tunable nature of the optimization, allowing for a customizable cost function and expert-specified restrictions placed on connections reducing the of the solution space. As a result, the simulated annealing framework notably resulted in a less complex network than the TAN technique and did not suffer from illogical connections that other models encountered. Conclusion Simulated annealing-based optimization resulted in BN topologies with comparable predictive accuracies to other out-of- box optimization techniques. The simulated annealing approach constructed more interpretable and logical topologies while allowing for customization within the cost function and initialization settings to allow for the inclusion of clinical knowledge in the optimization process. Although model accuracy is currently limited, this technique provides a preliminary approach for late toxicity prediction, which remains a complex task with no standard methodology for identifying at-risk patients. L. Boldrini 1 , G. Chiloiro 2 , D. Cusumano 3 , P. Yadav 4 , G. Yu 5 , A. Romano 2 , A. Piras 6 , L. Placidi 7 , S. Broggi 8 , F. Catucci 3 , J. Lenkowicz 7 , L. Indovina 7 , M.F. Bassetti 9 , Y. Yang 10 , C. Fiorino 8 , V. Valentini 2 , M.A. Gambacorta 2 1 Fondazione Policlinico Universitario "A. Gemelli" IRCCS , Radiology, Radiation Oncology and Hematology, Rome, Italy; 2 Fondazione Policlinico Universitario “A. Gemelli” IRCCS, Radiology, Radiation Oncology and Hematology, Rome, Italy; 3 Mater Olbia Hospital, Radiotherapy, Olbia, Italy; 4 Northwestern Memorial Hospital, Radiation Oncology, Chicago, USA; 5 University of California Los Angeles, Radiological Sciences, Los Angeles, USA; 6 Villa Santa Teresa, UO Radioterapia Oncologica, Bagheria, Italy; 7 Fondazione Policlinico Universitario “A.Gemelli” IRCCS, Radiology, Radiation Oncology and Hematology, Rome, Italy; 8 San Raffaele Scientific Institute, Medical Physics, Milan, Italy; 9 University of Wisconsin , Department of Human Oncology School of Medicine and Public Heath, Wisconsin , USA; 10 University of California Los Angeles, Radiological Sciences, Los Angeles, USA Purpose or Objective Early Regression Index (ERITCP) is able to efficaciously predict pathological complete response (pCR) on magnetic resonance (MR) images of patients with Locally Advanced Rectal Cancer (LARC). This study aims to investigate the potential role of Radiomics in adding discriminative power to ERITCP in patients undergoing neoadjuvant chemoradiotherapy on a 0.35 T MR Linac. Materials and Methods LARC patients (cT2-4 and/or cN0-2, cM0) treated with MR-guided Radiotherapy (MRgRT) with different schedules were retrospectively enrolled from three international institutions and divided in training and validation set for a pCR predictive model. PO-2090 Can Radiomics support Early Regression Index in predicting rectal cancer response to MRgRT?

Biologically effective dose (BED) conversion was used to compensate for the different RT schemes. 0.35T T2*/T1- weighted MR images were acquired during simulation and prior to each treatment fraction.

The rectal Gross Tumor Volume (GTV) was manually contoured on the 0.35 T MR images corresponding to the BED levels of 0,13,26,40,53 and 59 Gy and radiomics analysis was performed using MODDICOM, a dedicated image analysis platform compliant with the IBSI recommendations. Multiple logistic regression models, combining ERITCP with radiomic features, were calculated and compared with the model obtained using ERITCP as unique variable. Predictive performances were evaluated using Receiver Operating Characteristic (ROC) curve and model comparison was performed using Delong test. Results were evaluated on a separated validation set.

Results