MRI-derived prostate volume assessment: manual segmentation versus two-axis ellipsoid approximation and its implications for PSA density

Abstract Purpose To evaluate how accurately prostate volume estimated with the ellipsoid formula matches manual segmentation (reference standard) and to assess whether an individualized, volume-dependent adjustment of the ellipsoid weighting factor improves performance. Methods In this prospective study, patients with suspected prostate cancer (PCa) who underwent 3 T multiparametric MRI (T1WI, T2WI, DWI, DCE) between January 2022 and June 2023 were included. PSA values were obtained at imaging or within the preceding 4 weeks. Reference prostate volumes were derived from manual segmentation of all axial T2-weighted slices. Ellipsoid-based volumes were calculated per PI-RADS recommendations from sagittal and axial T2-weighted images using A × B × C × 0.52. Results Overall, 328 patients were analyzed (median age 68 years), including 196 without and 132 with PCa. Ellipsoid-based volumes showed excellent agreement with manual segmentation (ICC = 0.985; mean absolute difference 5.5 cc). Differences were larger in prostates > 80 cc, where the ellipsoid method tended to underestimate volume. Optimal volume-dependent weighting factors ranged from 0.50 to 0.54, but statistical modeling with variable weighting factors did not significantly improve model performance. For PSAd cut-offs of 0.1, 0.15, and 0.2 ng/mL/cc, the ellipsoid approximation achieved sensitivities of 0.86–0.95 and specificities of 0.88–0.97 compared with manual segmentation. Conclusion Manual segmentation and ellipsoid-based approximation show good agreement, underscoring the practical utility of the ellipsoid approximation for routine prostate volume assessment. While inaccuracies increase for larger glands, individualized weighting offers no meaningful benefit, and diagnostic performance for PSA density remains largely unaffected. Key Points Question Does the fast PI-RADS ellipsoid method estimate MRI prostate volume accurately versus manual segmentation, and do errors affect PSA density decisions? Findings In 328 men, ellipsoid volumes correlated strongly with segmentation (ρ = 0.98), underestimated larger glands (> 80 cc), and coefficient tuning added no meaningful improvement . Clinical relevance Ellipsoid-based volumetry provides a rapid, reliable basis for mpMRI prostate and PSA density assessment in routine clinical practice, supporting biopsy and active-surveillance decisions without time-consuming segmentation, while recognizing slightly higher error in very large prostates .