ESTRO 2022 - Abstract Book

S1411

Abstract book

ESTRO 2022

patients. The paediatric test cohort remained unchanged among the different methods and included 40 children. The Mean Absolute Error (MAE) was used to evaluate sCTs and as loss function of the network. Early stopping on the validation set was used as stopping criterion. Wilcoxon tests were performed to assess the significance of the observed differences with

a threshold value of 5%.

Results The average MAE within heads were respectively equal to 107 ± 20 HU, 143 ± 20 HU, 106 ± 19 HU and 102 ± 19 HU for (1_RSP), (2), (3) and (4). Training with RSP rather than HU in method (1) showed an improvement in average MAE within heads from 121 ± 22 HU to 107 ± 20 HU. Wilcoxon tests showed that these differences were significant (p<0.0001), except between methods (1_RSP) and (3) (p >0.25).

Conclusion Our analysis confirms the difficulty of generating paediatric sCTs directly from an adult model. The transfer learning method combined with a transformation of the scans into RSPs proposed in this study is an effective strategy to overcome the lack of patients in paediatric cohorts, and is extendable to adult patients. Dosimetric differences resulting from the different strategies have to be quantified in a near future.

PO-1622 Target definition for cardiac radioablation of ventricular tachycardia: A multimodal workflow

L. Rigal 1 , R. Martins 2 , K. Benali 3 , J. Bellec 4 , M. Lederlin 5 , R. De Crevoisier 6 , A. Simon 1

1 Université Rennes 1, LTSI - Inserm 1099, Rennes, France; 2 CHU Rennes, Cardiology Department, Rennes, France; 3 Saint- Etienne University Hospital, Cardiology Department, Saint Priest en Jarez, France; 4 CLCC Eugène Marquis, Medical Physics Department, Rennes, France; 5 CHU Rennes, Radiology and Medical Imaging Department, Rennes, France; 6 CLCC Eugène Marquis, Radiation Therapy Department, Rennes, France Purpose or Objective Precise target definition is a keystone of the promising cardiac radioablation (CR) technique for the treatment of ventricular tachycardia (VT). This step of treatment planning relies on multimodal data integration (cardiac CT scans, electro- anatomical mapping (EAM), PET...), as many anatomic and functional information must be exploited to locate the arrhythmogenic substrate. CR target definition is a difficult task, and prone to imprecisions due to the numerous data integration processes and to visualization limitations. The objective of this work was to propose a workflow for multimodal data integration to improve the robustness of CR target definition. Materials and Methods A target definition workflow was developed to generate a 3D mesh on which were fused all descriptors of interest extracted from multimodal images of a given patient. The left ventricle and myocardium were automatically segmented from cardiac Computed Tomography (cCT) image using a deep learning approach. The 3D mesh of the left ventricle was then built using the marching cubes algorithm. Myocardium thickness was computed on each point of the cCT mesh, as its distance to the closest non-ventricle, non-myocardium point.