Related Experiment Video
Updated: May 6, 2026
12:30
A Comprehensive Protocol for Manual Segmentation of the Medial Temporal Lobe Structures
Published on: July 2, 2014
19.7K
Fully automated prostate segmentation on MRI: comparison with manual segmentation methods and specimen volumes
Baris Turkbey1, Sergei V Fotin, Robert J Huang
11 Molecular Imaging Program, National Cancer Institute, National Institutes of Health, 10 Center Dr, MSC 1182, Bldg 10, Rm B69, Bethesda, MD 20892-1088.
AJR. American Journal of Roentgenology
|October 24, 2013
Summary
A new automated segmentation tool accurately estimates prostate volume from MRI scans, matching manual segmentation results. This 10-second process aids in prostate cancer diagnosis and treatment planning.
Area of Science:
- Radiology
- Medical Imaging
- Urology
Background:
- Accurate prostate volume estimation is crucial for prostate cancer diagnosis and treatment.
- Current methods like manual segmentation can be time-consuming.
Purpose of the Study:
- To compare prostate volume calculations from manual segmentation, automated segmentation, and ellipsoid formula against prostatectomy specimens.
- To evaluate the accuracy and speed of a novel automated segmentation system.
Main Methods:
- T2-weighted triplane MRI was performed on 98 patients.
- Prostate volumes were calculated using ellipsoid formula, manual segmentation, and an automated system.
- Results were validated against prostatectomy specimen volumes using correlation and Dice similarity coefficients.
Main Results:
- Strong positive correlations were found between prostatectomy volumes and both manual (R=0.89-0.91) and automated (R=0.88-0.91) segmentation.
- Automated segmentation achieved high Dice similarity coefficients (0.92 partial, 0.89 full).
- No significant difference was observed between manual and automated segmentation accuracy.
Conclusions:
- The automated 3D segmentation tool provides highly accurate prostate volume estimates.
- This tool operates within a clinically relevant timeframe of 10 seconds.
- It has potential applications in routine volume estimation, image registration, biopsy guidance, and decision support.
