Inter-observer variation in delineation of bladder and rectum contours for brachytherapy of cervical cancer

Background and purpose: In 3D treatment planning of low dose rate brachytherapy of cervical carcinoma the dose in bladder and rectum can be estimated from dose-volume histograms (DVHs). In this study, the influence of inter-observer variation in delineation of bladder and rectum on DVHs and dose at specific bladder and rectum points was investigated. Materials and methods: Three observers delineated bladder and rectum on axial CT images of ten patients. The highest minimum dose in bladder and rectum was determined for, respectively, 2 cm(3) (D-2) and 5 cm(3) (D-5)r as well as the dose at specific points placed on the bladder and rectum wall. Results: The inter-observer variation in D-2 was 10% (1 average relative SD) in bladder and 11% (1 SD) in rectum. In D-2 the variation was 8% (1 SD) in bladder and 11% in rectum. The variation in the bladder point was 13% (1 SD) and in the rectum point 11% (1 SD). Differences in delineation among the observers were caused by unclear organ boundaries on the CT images. Conclusions: Taking the inter-observer variation caused by delineation differences into account, dose in bladder and rectum can be determined within an accuracy of about 10% (1 SD). (C) 2000 Elsevier Science Ireland Ltd. All rights reserved.