Prostate volume on computed tomography correlates well with magnetic resonance imaging measurements and is reproducible across rater training levels

Background Data are lacking for the accuracy of computed tomography (CT) in measuring prostate size, which can streamline care and prevent invasive procedures. We evaluate agreement and intra/inter-observer variability in prostate sizing between CT and magnetic resonance imaging (MRI) planimetry for a wide range of gland sizes. Methods We retrospectively reviewed 700 patients who underwent MRI fusion biopsy at a single institution and identified 89 patients that had a CT within 2 years of the MRI. Six reviewers from different training levels were categorized as student, resident, or attending and each measured prostate size on CT by the prolate ellipse method. Bland-Altman analysis determined the degree of agreement between CT and MRI. Inter- and intra-observer reliability was calculated for CT. Results Mean CT volume was higher than MRI volume in the < 60 g group (51.5 g vs. 44.5 g, p = 0.004), but not in the >= 60 g group (101 g vs. 100 g, p = 0.458). The bias for overestimation of prostate volume by CT was 4.1 g across prostate volumes, but the proportional agreement between modalities improved with size. The Pearson correlation coefficient between CT/MRI and inter/intra-rater reliability for CT increased in the >= 60 g vs. the < 60 g group for all training levels. Conclusions Our data show that there is greater clinical utility for prostate size estimation by CT than previously established, particularly for larger glands where accurate size estimation may influence therapeutic decisions. In larger glands, prostate size estimation by CT is also reproducible across various training levels.