Intra- and inter-observer variability in contouring prostate and seminal vesicles: implications for conformal treatment planning

Background and purpose: Accurate contouring of the clinical target volume (CTV) is a fundamental prerequisite for successful conformal radiotherapy of prostate cancer. The purpose of this study was to investigate intra-and inter-observer variability in contouring prostate (P) and seminal vesicles (SV) and its impact on conformal treatment planning in our working conditions. Materials and methods: Inter-observer variability was investigated by asking five well-trained radiotherapists of contouring on CT images the P and the SV of six supine-positioned patients previously treated with conformal techniques. Short-term intra-observer variability was assessed by asking the radiotherapists to contour the P and SV of one patient for a second time, just after the first contouring. The differences among the inserted volumes were considered for both intra-and inter-observer variability. Regarding intra-observer variability, the differences between the two inserted contours were estimated by taking the relative differences in correspondence to the CT slices on BEV plots (antero-posterior and left-right beams). Concerning inter-observer variability, the distances between the internal and external envelopes of the inserted contours (named projected diagnostic uncertainties or PDUs) and the distances from the mean inserted contours (named mean contour distances or MCDs) were measured from BEV plots (i.e. parallel to the CT slices). Results: Intra-observer variability was relatively small (the average percentage variation of the volume was similar to 5%; SD of the differences measured on BEV plots within 1.8 mm). Concerning inter-observer variability, the percentage SD of the inserted volumes ranged from 10 to 18%. Differences equal to 1 cm in the cranio-caudal extension of P + SV were found in four out of six patients. The largest interobserver variability was found when considering the anterior margin in the left-right beam of P top (MCD = 7.1 mm, 1 SD). Relatively high values for MCDs were also found for P bottom, for the posterior and lateral margins of P top (2.6 and 3.1 mm, respectively, 1 SD) and for the anterior margin of SV (2.8 mm, 1 SD). Relatively small values were found for P central (from 1.4 to 2.0 mm, 1 SD) and the posterior margin of SV (1.5 mm, 1 SD). Conclusions: The application of larger margins taking inter-observer variability into account should be taken into consideration for the anterior and the lateral margins of SV and P top and for the lateral margin of P. The impact of short-term intra-observer variability does not seem to be relevant. (C) 1998 Elsevier Science Ireland Ltd. All rights reserved.