Related Experiment Video
Updated: Jun 1, 2026
05:33
Three-Dimensional Shape Modeling and Analysis of Brain Structures
Published on: November 14, 2019
Statistical shape and texture model of quadrature phase information for prostate segmentation
Soumya Ghose1, Arnau Oliver, Robert Martí
1Computer Vision and Robotics Group, University of Girona, Girona, Spain. soumyaghose@gmail.com
Summary
This study presents an efficient and accurate method for segmenting the prostate in transrectal ultrasound (TRUS) images, improving diagnosis and treatment for prostate conditions.
Area of Science:
- Medical Imaging
- Computational Anatomy
- Biomedical Engineering
Background:
- Accurate prostate segmentation in transrectal ultrasound (TRUS) images is crucial for diagnosing and treating prostate conditions like hypertrophy and cancer.
- Challenges in TRUS image segmentation include low signal-to-noise ratio, speckle noise, calcifications, and intensity variations within the prostate.
- Existing methods often struggle with computational efficiency and accuracy in complex imaging scenarios.
Purpose of the Study:
- To develop a computationally efficient and accurate computer-aided method for prostate segmentation in TRUS images.
- To address the challenges posed by noise and intensity variations in TRUS imaging for precise prostate delineation.
- To enhance the diagnostic and therapeutic capabilities through improved prostate volume estimation.
Main Methods:
- A multi-resolution framework incorporating texture features within a parametric deformable statistical model of shape and appearance was employed.
- Log-Gabor quadrature filters extracted local phase information for robust texture representation, invariant to gray-level shifts and unaffected by noise.
- A parametric contour model, derived from principal component analysis of prior shape and texture data, was optimized for segmentation.
Main Results:
- The proposed method achieved a high mean Dice similarity coefficient of 0.95 ± 0.02.
- A low mean absolute distance of 1.26 ± 0.51 mm was obtained, indicating precise boundary delineation.
- Validation using 24 TRUS images across 6 datasets demonstrated robust performance in a leave-one-patient-out framework.
Conclusions:
- The developed method offers a computationally efficient solution for prostate segmentation in TRUS images.
- It provides accurate segmentation results, effectively handling intensity heterogeneities and imaging artifacts.
- This approach holds significant potential for improving clinical diagnosis and treatment planning for prostate diseases.
