ESTRO 2023 - Abstract Book

S1383

Digital Posters

ESTRO 2023

(p-value <= 0.0001, given by a signed-rank Wilcoxon test between the ConvNext method and the VGG method) were obtained when using the ConvNext-based perceptual loss.

Conclusion The use of ConvNext-based perceptual loss results in a significant visual and quantitative improvement in unsupervised sCT generation. Our work shows: (1) The feasibility of training with non-registered data using perceptual loss (2) an alternative to VGG for the perceptual loss must continue to be studied.

PO-1678 DirectDensity CT image reconstruction algorithm in clinical practice: a survey

B. van der Heyden 1 , P. Wohlfahrt 2 , V. Taasti 3 , E. Bogaert 4

1 Ghent University, Department of Electronics and Information Systems, Gent, Belgium; 2 Massachusetts General Hospital and Harvard Medical School, Department of Radiation Oncology, Boston, USA; 3 Maastricht University Medical Centre+, Department of Radiation Oncology (Maastro), Maastricht, The Netherlands; 4 Ghent University Hospital, Department of Radiotherapy-Oncology, Gent, Belgium Purpose or Objective The DirectDensity™ (DD) algorithm reconstructs CT images interpretable as electron or mass density independent of X-ray spectrum. Hence, the tube potential can be selected automatically to improve patient-specific image contrast without the need of several energy-dependent calibration curves in the treatment planning system (TPS). No protocols on the optimal use of DD exist in radiotherapy, so this work assessed the current clinical use of DD and aims to provide guidelines. Materials and Methods An online survey was conducted to investigate the DD usage in centres that have implemented it clinically or showed interest in (near) future. The questionnaire was sent to 30 radiotherapy centres with known DD availability. Results The survey was exclusively answered by medical physicists and had a 77% response rate (2US, 21EU centres). DD was implemented combined with automatic tube voltage selection software (CARE kV) in ten centres. Six of these centres avoided 70kV due to increased CT number instabilities and image noise, while one centre also set kV restrictions ( ≥ 100kV) for breast/lung cancer treatments. A main reason to use DD was the ability to select variable tube settings for optimal