Artificial intelligence-generated targets and inter-observer variation in online adaptive radiotherapy of bladder cancer.

Background And Purpose: Daily target re-delineation in online adaptive radiotherapy (oART) introduces uncertainty. The aim of this study was to evaluate artificial intelligence (AI) generated contours and inter-observer target variation among radiotherapy technicians in cone-beam CT (CBCT) guided oART of bladder cancer.

Materials And Methods: For each of 10 consecutive patients treated with oART for bladder cancer, one CBCT was randomly selected and retrospectively included. The bladder (CTV-T) was AI-segmented (CTV-T). Seven radiotherapy technicians independently reviewed and edited CTV-T, generating CTV-T. Contours were benchmarked against a ground truth contour (CTV-T) delineated blindly from scratch. CTV-T and CTV-T were compared to CTV-T using volume, dice similarity coefficient, and bidirectional local distance. Dose coverage (D>95 %) of CTV-T was evaluated for treatment plans optimized for CTV-T and CTV-T with clinical margins. Inter-observer variation among CTV-T was assessed using coefficient of variation and generalized conformity index.

Results: CTV-T ranged from 48.7 cm to 211.6 cm. The median [range] volume difference was 4.5 [-17.8, 42.4] cm for CTV-T and -15.5 [-54.2, 4.3] cm for CTV-T, compared to CTV-T. Corresponding dice similarity coefficients were 0.87 [0.71, 0.95] and 0.84 [0.64, 0.95]. CTV-T was adequately covered in 68/70 plans optimized on CTV-T and in 6/10 plans optimized on CTV-T with clinical margins. The median [range] coefficient of variation was 0.08 [0.05, 0.11] and generalized conformity index was 0.78 [0.71, 0.88] among CTV-T.

Conclusions: Target re-delineation in CBCT-guided oART of bladder cancer demonstrated non-isotropic inter-observer variation. Manual adjustment of AI-generated contours was necessary to cover ground truth targets.