ESTRO 2022 - Abstract Book

S1254

Abstract book

ESTRO 2022

data was randomly split into 70%, 20% and 10% groups for training, validation and testing respectively. An input series lengths of 5, the number obtained in one week, along with the date and time at which the scans were obtained, were presented to the network and predictions of the next 10 scans were produced. The accuracy of predictions was evaluated using mean squared error (MSE) to compare predicted scans to ground truths. Results The average MSE between the network predictions (Fig. 1C) and ground truths (Fig. 1B) in a test set of 27 patients were 2.79x10 3 for predictions of anatomy made 0-24 hours after the final input scan, 3.85x10 3 for 48-72 hours after, 6.54x10 3 for 144-168 hours after and 1.36x10 4 for 312-336 hours after. However, only 11 of the 27 series in the test set contained 15 or more scans. The 144-168 hours and 312-336 hours results are averages of 22 and 11 prediction accuracies respectively.

Conclusion We produced a CRNN architecture for predicting changes in head and neck patient anatomy given one week of patient scans. Predictions of anatomy within 72 hours of the final input image exhibit the modelling capabilities of the network. Further work will focus on improving the model accuracy to provide useful anatomical predictions for one week in the future and beyond, allowing for its use as a decisions making tool, improving resource allocation.

PO-1479 SGRT setup patient accuracy in breast cancer patients compared to two different IGRT workflows

C. Anson 1 , N. Jornet 1 , N. Ventosa 2 , S. Bermejo 2 , J. Pérez-Alija 1 , P. Gallego 1 , A. Latorre-Musoll 1 , N. García 1 , H. Vivancos 1 , M. Barceló 1 , A. Ruiz 1 , F. Leo 1 , P. Carrasco 1 1 Hospital de la Santa Creu i Sant Pau, Medical Physics, Barcelona, Spain; 2 Hospital de la Santa Creu i Sant Pau, Radiation Oncology, Barcelona, Spain Purpose or Objective Surface guided radiotherapy (SGRT) is reported to improve patient setup for whole-breast radiotherapy. The purpose of this study is to evaluate the patient SGRT accuracy for pre-imaging setup considering two different IGRT workflows: orthogonal kV-MV images and CBCT guidance. Materials and Methods Patient setup accuracy during the first treatment session of 31 breast cancer patients was retrospectively evaluated. The workflow was performed in a TrueBeam unit and the optical SGRT system Align-RT (Vision-RT) as follows: • First, patients were positioned by aligning tattoos to the room lasers and TPS isocentre shifts were applied to the couch. After this, patient arms and breast surface were matched to the simulation position with the aid of Vision RT postural video license and a ROI including the breast. From this, system delta couch shifts were performed only for the three directions: vertical, lateral and longitudinal, and recorded as ‘SGRT’ deviation. • Orthogonal kV-MV images guidance were acquired. The register was done to bony structures (rib bone and sternum). Differences between initial (pre-imaging) and treatment couch positions were then recorded as ‘KV- MV’ deviation. • Immediately after, a CBCT was performed and an automatic image registration to the PTV structure was done. Differences between initial (pre-imaging) and treatment couch positions were then recorded as ‘CBCT’ deviation. If the deviations either with kV-MV or CBCT images were within 0.5 cm tolerance, no table shift was applied.