Prostate positioning using cone-beam computer tomography based on manual soft-tissue registration: Interobserver agreement between radiation oncologists and therapists

Purpose: To check the interobserver agreement between radiation oncologists and therapists (RTT) using an on- and off-line cone-beam computer tomography (CBCT) protocol for setup verification in the radiotherapy of prostate cancer. Materials and methods: The CBCT data from six prostate cancer patients treated with hypofractionated intensity-modulated radiotherapy (IMRT) were independently reviewed off-line by four observers (one radiation oncologist, one junior and two senior RTTs) and benchmarked with on-line CBCT positioning performed by a radiation oncologist immediately prior to treatment. CBCT positioning was based on manual soft-tissue registration. Agreement between observers was evaluated using weighted Cohen's kappa statistics. Results: In total, 152 CBCT-based prostate positioning procedures were reviewed by each observer. The mean (± standard deviation) of the differences between off- and on-line CBCT-simCT registration translations along the three directions (antero-posterior, latero-lateral and cranio-caudal) and rotation around the antero-posterior axis were - 0.7 (3.6) mm, 1.9 (2.7) mm, 0.9 (3.6) mm and - 1.8 (5.0) degrees, respectively. Satisfactory interobserver agreement was found, being substantial (weighted kappa > 0.6) in 10 of 16 comparisons and moderate (0.41-0.60) in the remaining six comparisons. Conclusions: CBCT interpretation performed by RTTs is comparable to that of radiation oncologists. Our study might be helpful in the quality assurance of radiotherapy and the optimization of competencies. Further investigation should include larger sample sizes, a greater number of observers and validated methodology in order to assess interobserver variability and its impact on high-precision prostate cancer IGRT. In the future, it should enable the wider implementation of complex and evolving radiotherapy technologies.