ESTRO 2023 - Abstract Book

S1692

Digital Posters

ESTRO 2023

the registration of the test images was 0.43 mm. The uncertainty analysis showed that our method correctly estimated the model uncertainty while underestimated the image uncertainty. Conclusion In summary, a fast and accurate DIR method which predict the DDF uncertainty was developed using CNNs and drop-out layers. It was shown that the accuracy of the method to generate the daily structures was not statistically different to Elastix. The uncertainty analysis has shown that this network was able to estimate the model uncertainty but underestimates the image uncertainty. Future works must improve DDF uncertainty prediction and evaluation to facilitate physician’s review of the structures and save time in a daily on-line adaptive workflow.

PO-1937 Planning target margins for CBCT guided radiotherapy of the breast/thoracic wall and lymph nodes

M. Falk 1 , M. Hedman 2,3 , M. Bolin 1 , J. Falk 2 , K. Karlsson 1,3 , E. Onjukka 1,3 , B. Sorcini 1 , G. Gagliardi 1,3

1 Karolinska University Hospital, Section of Radiotherapy and Engineering, Medical Radiation Physics and Nuclear Medicine, Stockholm, Sweden; 2 Karolinska University Hospital, Department of Radiation Oncology, Stockholm, Sweden; 3 Karolinska Institutet, Department of Oncology-Pathology, Stockholm, Sweden Purpose or Objective The purpose of this study was to determine appropriate CTV to PTV margins for patients treated with adjuvant CBCT guided radiotherapy of the breast/thoracic wall and lymph nodes. Materials and Methods A total of 72 patients were enrolled in the study. Five of these were excluded as they had potential need of adaptation of their treatment plans due to set-up errors outside our clinics guidelines (7mm). Of the remaining 67 patients, 32 were treated in free breathing (FB) and 35 in deep inspiration breath-hold (DIBH). For three of the treatment fractions, CBCTs were acquired before and after the treatment and the setup uncertainties were evaluated by comparing the patient position at the time of treatment to that at the time of the CT acquisition for treatment planning. The intra-fractional uncertainties were evaluated by comparing images before and after the treatment. In addition, MV-imaging of a tangential treatment field was used to evaluate the setup uncertainties in the longitudinal (long) direction of the caudal part of the target as this anatomy was not visualized in the CBCT-image. Nine different anatomy sites were evaluated; (A) the proximal part of the clavicular bone next to the sternoclavicular joint, (B) the manubriosternal joint, (C) the middle of the sternum, (D) the proximal part of the vertebra C7, (E) the proximal coracoid process, (F) the proximal infraglenoid tubercle, (G) the proximal thoracic wall at the level of the manubriosternal joint, (H) the proximal thoracic wall at the level of the middle of the sternum and (I) the rib proximal to the caudal part of the target. Five patients received treatment of only the breast/thoracic wall and their images did not include the sites D-F. The image analysis was done in the Offline Review application of Aria 16 (Varian Medical Systems) using an automatic tool for matching with a VOI of 1.5 cm in the cranio caudal (CC) direction for site A-H. Matching included translation in 3D, except for G-H where matching was done in vertical (vert) and lateral (lat) directions only, as these structures were too large to be properly identified in the long direction using a VOI of 1.5 cm in the CC direction. The MV-images were analyzed using manual match. Margins were calculated using the Van Herk margin formula ensuring 90% of patients receiving 95% of the prescribed dose. Statistical analysis of the difference in uncertainties in set-up and intra fraction motion for patients treated in FB and DIBH was done using a two sided students t-test. Results For some anatomy sites and directions, there was no significant difference between the uncertainties in FB and DIBH treatments and a joint margin was calculated (table 1 and figure 1). The largest margins were calculated for D-F in the long and vrt directions.