EUVIMEDOpdag EUVIMED
VergleichsstudieOpen access · CC BY 4.0

Defining ground truth for prostate segmentation of transrectal ultrasound images: Inter‐ and intra‐observer variability of manual versus semi‐automatic methods

Louis Lenfant, Clément Beitone, Jocelyne Troccaz, Gaelle Fiard, Bernard Malavaud, Sandrine Voros, Pierre C Mozer

Medical Physics · 2025-08 · 52 (8)

DOI: 10.1002/mp.18025 ↗

At a glance

Vergleichsstudie in Medical Physics, published online on 2025-08-11. Written by Louis Lenfant and 6 co-authors. Institution: Predictive Onco-Urology, GRC n°5, Sorbonne Université, AP-HP, Hôpital Pitié-Salpêtrière, Urology, Paris, France. Major MEDLINE topics: Prostate, Rectum, Image Processing, Computer-Assisted. Cited 2 times so far (Crossref); 28 references. The full text is freely available (open access, CC BY 4.0). Funded by: French Public Bank of Investments, Koelis, Vermon.

AbstractSubjectsBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Background

Accurate prostate segmentation in transrectal ultrasound (TRUS) imaging is essential for diagnosis, treatment planning, and developing artificial intelligence (AI) algorithms. Although manual segmentation is often recommended as the ground truth for AI training, it is time‐consuming, prone to inter‐ and intra‐observer variability, and rarely used in everyday clinical practice. Semi‐automatic methods provide a faster alternative but lack thorough multi‐operator evaluations. Understanding variability in segmentation methods is crucial to defining a reliable reference standard for future AI training.

Purpose

To investigate the inter‐individual variability in manual and semi‐automatic prostate contour segmentation on 3D TRUS images and to compare both approaches to determine the most consistent method that could serve as a reference standard for future AI model development.

Methods

This study is a methodological investigation and not an AI study. Four urology experts independently performed manual and semi‐automatic segmentation on 100 prostate 3D TRUS exams obtained from patients undergoing fusion prostate biopsy. Inter‐individual and intra‐individual variability for manual segmentation was assessed using the Average Surface Distance (ASD) between manually placed points and a reference mesh. Two methods were used to create the reference prostate mesh after manual point positioning: a statistical shape model (manual_SSM) and a deformable model (manual_soft‐SSM). Semi‐automatic segmentations were evaluated using ASD, Dice similarity coefficient, and Hausdorff distance. A Simultaneous Truth and Performance Level Estimation (STAPLE) like consensus method was applied to assess variability across experts in semi‐automatic segmentation. Statistical comparisons used Wilcoxon tests, and effect sizes were calculated using Cohen's d. Bonferroni correction was applied for multiple comparisons. A significance level of p < 0.05 (adjusted as needed) was used.

Results

Manual segmentation inter‐individual variability was higher with the manual_SSM method [ASD = 2.6 mm (Inter Quartile Range (IQR) 2.3–3.0)] compared to the manual_soft‐SSM [ASD = 1.5 mm (IQR 1.2–1.8), P < 0.001]. Intra‐individual variability also showed lower ASD values with manual_soft‐SSM compared to manual_SSM, [(1.0 (0.8‐1.1) versus 2.2 (1.9‐2.6), p < 0.001], respectively. For semi‐automatic segmentation, inter‐individual variability yielded an ASD of 1.4 mm (IQR 1.1–1.9), Dice of 0.90 (IQR 0.88–0.92), and Hausdorff distance of 5.7 mm (IQR 4.47–7.36). Manual and semi‐automatic segmentation comparisons demonstrated an ASD of 1.43 mm (IQR 1.20–1.90).

Conclusions

The semi‐automatic segmentation method evaluated in this study demonstrated comparable accuracy to manual segmentation while reducing inter‐ and intra‐individual variability. These findings suggest that the tested semi‐automatic approach can serve as a reliable reference standard for AI training in prostate segmentation.

Subjects

MeSH terms (MEDLINE)

ProstateRectumImage Processing, Computer-AssistedHumansMaleUltrasonographyObserver VariationAutomationProstatic NeoplasmsImaging, Three-Dimensional

Author keywords

*artificial intelligencealgorithmsimage interpretation, computer‐assisted/*methodsimage‐guided biopsy*imaging, three‐dimensional*/methodsimaging, three‐dimensional/*methodspattern recognition, automated/*methodsprostatic neoplasms*/diagnostic imagingreproducibility of results

Authors and institutions

  • Predictive Onco-Urology, GRC n°5, Sorbonne Université, AP-HP, Hôpital Pitié-Salpêtrière, Urology, Paris, France.
  • Université Grenoble Alpes, CNRS, INSERM, Grenoble INP, TIMC, Grenoble, France.
  • Institut des Systèmes Intelligents et Robotique (ISIR), Sorbonne Université, CNRS UMR 7222, INSERM U1150, Paris, France.
  • Department of Urology, Université Grenoble Alpes, CNRS, CHU Grenoble Alpes, Grenoble INP, TIMC, Grenoble, France.
  • Department of Urology, Institut Universitaire du Cancer Toulouse-Oncopôle, Toulouse, France.

Funding

  • French Public Bank of Investments
  • Koelis
  • Vermon

Cite

APA

Lenfant, L., Beitone, C., Troccaz, J., Fiard, G., Malavaud, B., Voros, S., & Mozer, P. C. (2025). Defining ground truth for prostate segmentation of transrectal ultrasound images: Inter‐ and intra‐observer variability of manual versus semi‐automatic methods. Medical Physics, 52(8). https://doi.org/10.1002/mp.18025

Vancouver

Lenfant L, Beitone C, Troccaz J, Fiard G, Malavaud B, Voros S, et al. Defining ground truth for prostate segmentation of transrectal ultrasound images: Inter‐ and intra‐observer variability of manual versus semi‐automatic methods. Medical Physics. 2025;52(8). doi:10.1002/mp.18025

BibTeX
@article{Lenfant2025,
  author = {Lenfant, Louis and Beitone, Clément and Troccaz, Jocelyne and Fiard, Gaelle and Malavaud, Bernard and Voros, Sandrine and Mozer, Pierre C},
  title = {Defining ground truth for prostate segmentation of transrectal ultrasound images: Inter‐ and intra‐observer variability of manual versus semi‐automatic methods},
  journal = {Medical Physics},
  year = {2025},
  volume = {52},
  number = {8},
  doi = {10.1002/mp.18025}
}
Start for free →