Computed tomography determination of prostate volume and maximum dimensions: a study of interobserver variability

Objective: (1) To evaluate the reproducibility of prostate volume, maximum dimensions and geometrical center coordinates determination using computed tomography (CT) and (2) to identify patterns of interobserver variability. Materials and methods: Forty patients, suitable for our brachytherapy program, were selected for the study. All patients underwent CT scanning and the prostate volumes were determined by three radiation oncologists. Measurements of geometrical center coordinates, maximum organ dimensions in the anterior-posterior (AP), lateral (Lat) and longitudinal (Long) axes as well as prostate volumes were recorded. This yielded 840 measurements of seven variables for analysis. The means and corresponding standard deviations (SD) of each variable were calculated for each patient. The SDs were then averaged and presented as indices of dispersion. Average variations from the mean were also calculated for each observer along with the SDs. Results: Analysis of the geometrical center coordinates revealed acceptable variability amongst observers. For the AP, Lat and Long coordinates the SDs were 0.78, 0.89 and 1.72 mm, respectively. The corresponding values for the maximum organ dimensions were 2.54, 2.72 and 4.43 mm, respectively. While the volumes outlined by observer B were less than or equal to the mean in 95% of cases and those of observer C were greater than or equal to the mean in 93% of cases, the volumes of observer A were equally distributed above and below the mean (48% in both cases). Conclusion: The determination of the geometrical center coordinates was reproducible amongst observers. The largest variations were seen with the Long axis. The volume determination is more variable. However, a characteristic trend was seen amongst observers when their volumes were compared to the mean volumes of the group. (C) 2002 Elsevier Science Ireland Ltd. All rights reserved.