Intra- and inter-observer variability in contouring prostate and seminal vesicles: implications for conformal treatment planning
Introduction
Many sources of possible errors in planning and delivering conformal radiotherapy for prostate cancer have been investigated and their impact on treatment planning (i.e. the choice of appropriate margins in defining the planning target volume (PTV) [11]) has been widely analyzed 1, 3, 4, 11, 15, 22, 24, 29. However, even though accurate contouring of the clinical target volume (CTV) is a fundamental prerequisite for successful conformal radiotherapy, practically no literature exists about inter- and intra-observer variability in contouring prostate and seminal vesicles, which, in many cases, correspond to the CTV in the treatment of localized prostate cancer.
The reason for this lack of knowledge should depend on a generally accepted opinion that, in most cases, the prostate and the seminal vesicles should be well visible on CT images. However, as indicated by some authors, an estimate of intra- and inter-doctor variability is important 13, 29, due to the reduced margins which are applied during the beam's-eye-view (BEV)-based conformal shaping of the beams. Both intra- and inter-observer variations could also be influenced by technical aspects, such as the image resolution, the choice of grey levels, the thickness of CT slices and the distance between them and the use of contrast liquids in the bladder and/or rectum.
The aim of this current work is to investigate intra- and inter-observer variability in contouring the prostate and seminal vesicles in our routine working conditions.
Five well-trained radiotherapists were asked to contour the prostate and the seminal vesicles of six supine-positioned patients previously treated with conformal radiotherapy for prostate cancer. For one patient, all doctors were asked to reinsert the same contour after a few minutes, in order to assess (short-term) intra-observer variability.
As an example of the impact of inter-observer variability on conformal treatment planning, only one patient was considered. The PTVs corresponding to the CTVs (prostate+seminal vesicles) contoured by each doctor were generated through automatic volume expansion. A four-field conformal technique was simulated for each PTV and the relative differences in dose statistics and dose–volume histograms (DVHs) of the rectum and the bladder were reported.
Section snippets
CT simulation: procedures and patient characteristics
Five well-trained radiotherapists with at least 8 years clinical experience and 3 years experience in 3D multi-slice CT treatment planning participated in this study. The CT images of six consecutive patients with localized prostate cancer, previously irradiated with conformally-shaped fields, were considered. This sample is representative of our daily working conditions. All patients had been submitted to a CT scan in the supine position (Toshiba TCT 900S) with their bladders full and with
Volumes
The percentage variations in the inserted volumes were relatively low, ranging from 1.5 to 9% (average 5%), among the different doctors.
BEV plot analysis
When considering the BEV plot analysis, the mean differences between the first and the second contours were about 0 with SD ranging from 0.8 to 1.8 mm (see Table 1) for the various positions/directions. Intra-observer variability was higher for the lateral direction (in the AP view) with respect to the posterior and anterior directions (in the lateral view).
Discussion
A number of studies have investigated the potential impact of intra- and inter-observer variability on contouring CTVs/PTVs and organs at risk (OARs) during the treatment planning process 5, 7, 8, 10, 13, 14, 18, 19, 27, 28, especially for head and neck and lung malignancies. Jones et al. [13]roughly estimated a standard deviation in the BEV margin of ~3 mm in the prostate. However, this estimate was based on the data which referred to only three doctors contouring P of one patient.
Intra-doctor
Conclusions
To our knowledge, this is the first extensive study which tries to assess the impact of intra- and inter-observer variability on contouring P+SV in clinically suitable conditions.
The most important result is that a relatively high inter-observer variability was found for P bottom and top and for anterior/lateral margins of SV. Inter-observer variability for the anterior margin of SV was found to decrease when contrast liquid was not added in the bladder. The need to use larger margins in the
References (30)
- Austin-Seymour, M., Kalet, I., McDonald, J., et al. Three dimensional planning target volumes: a model and a software...
- Algan, O., Hanks, G.E. and Shaer, A.H. Localization of the prostatic apex for radiation treatment planning. Int. J....
- Bel, A., van Herk, M. and Lebesque, J.V. Target margins for random geometrical treatment uncertainties in conformal...
- Bijhold, J., Lebesque, J.V., Hart, A.A.M. and Vijlbrief, R.E. Maximising set-up accuracy using portal images as applied...
- Cellai, E., Biti, G.P., Banci Buonamici, F., et al. The choice of PTV: a crucial point in the treatment plan...
- Cox, J.A., Zagoria, R.J. and Raben, M. Prostate cancer: comparison of retrograde urethrography and computed tomography...
- Dische, S., Williams, C.R. and Saunders, M.I. The definition of the tumor target volume – A further frontier for...
- Dowsett, R.J., Galvin, J.M., Cheng, E., et al. Contouring structures for 3-dimensional treatment planning. Int. J....
- Fiorino, C., Reni, M., Cattaneo, G.M., Bolognesi, A. and Calandrino, R. Comparing three, four and six-fields conformal...
- Hamilton, C.S., Denham, J.W., Joseph, D.J., et al. Treatment and planning decisions in non-small cell carcinoma of the...
Cited by (323)
NRG Oncology Assessment of Artificial Intelligence Deep Learning–Based Auto-segmentation for Radiation Therapy: Current Developments, Clinical Considerations, and Future Directions
2024, International Journal of Radiation Oncology Biology PhysicsPrototypical few-shot segmentation for cross-institution male pelvic structures with spatial registration
2023, Medical Image AnalysisReal-world validation of Artificial Intelligence-based Computed Tomography auto-contouring for prostate cancer radiotherapy planning
2023, Physics and Imaging in Radiation OncologyProspective imaging comparison of anatomic delineation with rapid kV cone beam CT on a novel ring gantry radiotherapy device
2023, Radiotherapy and Oncology