Reproducibility of prostate volume measurements from transrectal ultrasonography by an automated and a manual technique

Objective To assess the reproducibility of estimates of prostate volume determined by planimetry from transrectal ultrasonography (TRUS) images. Materials and methods Two sequential sessions of images obtained by TRUS were obtained from 30 patients, with the ultrasound probe removed and inserted between the sessions. The stored images were outlined, both manually and by computer, and measurements of prostate volume obtained planimetrically. In addition, the ability of the urologist to accurately draw the contour was assessed by outlining predefined contours. Results The mean (so) variability of manual outlining between sessions was 3.5 (3.4)%, within one session was 1.7 (1.3)% and of computer outlining between sessions was 4.3 (3.8)%. Comparing the results of manual and computer outlining showed a mean (so) variability of 7.5 (5.6)%, with larger values obtained from computer outlining. The mean (so) variation in manually outlining predefined contours was 1.4 (1.4)%. Conclusions The variability of computer outlining was slightly higher than expected theoretically. The within-session variability was higher than the variation caused by errors in outlining predefined contours, indicating that the interpretation of TRUS images differed with time. Automated determination of prostate volume can save time during clinical investigation and the variability is within the clinically acceptable range of 5%.