Purpose: To compare reconstructed volumes calculated via planimetry and frustum algorithms in the context of stepped transrectal ultrasound (US) imaging, and to estimate the reconstruction error for prostate volumes.
Methods: Prostate contours for 40 permanent implant patients were delineated on magnetic resonance (MR) and transrectal US images by a radiation oncologist. Simulated images of ellipsoid and truncated cone geometrical objects were constructed to determine volume calculation accuracy. Simulation results were used to deduce the algorithm-associated error made when calculating transrectal US prostate volumes.
Results: For imaging without deliberate slice positioning, planimetry reconstruction was mostly accurate while the frustum algorithm underestimated the volume. The discrepancy was mostly due to the end slice reconstruction. For slice positioning that reflected US image acquisition, planimetry overestimated by half the superior slice volume on average while frustum underestimated by half the inferior slice volume. The estimated algorithm errors for prostate contours were 4% and -3%, respectively.
Conclusions: The planimetry and frustum algorithms offer different interpretations for reconstruction and yield systematic differences in calculated volumes. Both algorithms introduce bias into transrectal US prostate volume determinations that may have clinical implications, planimetry overestimating and frustum underestimating the volume.