ESTRO 2024 - Abstract Book

S4029

Physics - Inter-fraction motion management and offline adaptive radiotherapy

ESTRO 2024

The RMSE and MAE values of CNN model reported 0.012 ± 0.001 and 0.010 ± 0.001, respectively. Figure 2 shows the absolute difference between the synthetic 4D-MRI and the XCAT-generated 4D-MRI, for each respiratory phase. For the testing dataset (n= 200), the SSIM and nRMSE values were 0.96 ± 0.06 and 0.22 ± 0.08, respectively. DC, VPD, and COMS were 0.92 ± 0.06, 3.08 ± 0.3.73 %, and 2.3 ± 2.1 mm, respectively for the target volume. The proposed method demonstrated robust results comparing the reference 4D-CBCT conditions ( ref ) to the different 4D-CBCT scenarios (SSIM ref = 0.93 ± 0.02, SSIM RP4 = 0.94 ± 0.02, SSIM RP8 = 0.93 ± 0.02, SSIM DA20 = 0.91 ± 0.03, SSIM DA40 = 0.91 ± 0.03, SSIM CW2.5 =0.91 ± 0.03, SSIM CW1 = 0.86 ± 0.05). Finally, the activated features of the CNN model were evaluated, confirming the recognition of important anatomical areas within the prediction process.

Conclusion: