Related Experiment Video
Updated: Nov 19, 2025
09:11
Use of MRI-ultrasound Fusion to Achieve Targeted Prostate Biopsy
Published on: April 9, 2019
22.0K
Machine learning and registration for automatic seed localization in 3D US images for prostate brachytherapy
Hatem Younes1, Jocelyne Troccaz1,2, Sandrine Voros1,2
1University of Grenoble Alpes, CNRS, TIMC-IMAG, F-38000, Grenoble, France.
Medical Physics
|January 29, 2021
Summary
This study presents a new method for accurately locating radioactive seeds in 3D ultrasound images, crucial for adaptive radiation therapy. The algorithm achieves high precision, improving treatment planning and delivery for brachytherapy patients.
Area of Science:
- Medical Imaging
- Radiation Oncology
- Biomedical Engineering
Background:
- Accurate radioactive seed localization is essential for advanced radiation therapy protocols like adaptive brachytherapy.
- Real-time ultrasound (US) imaging presents challenges for seed detection due to seed size and artifacts.
Purpose of the Study:
- To develop and evaluate a robust, automatic method for precise radioactive seed localization in 3D US images.
- To enable accurate comparison with planned seed positions for real-time dosimetric adjustments.
Main Methods:
- A prelocalization step identifies needles, focusing the search for seeds within a region of interest (ROI).
- Bayesian classification and Support Vector Machine (SVM) are used for ROI binarization and false positive removal.
- Iterative Closest Point (ICP) and Sum of Squared Differences (SSD) algorithms refine seed and strand localization.
Main Results:
- The method achieved mean localization errors of 1.09 ± 0.61 mm on phantom images.
- On clinical images, mean localization errors were 1.44 ± 0.45 mm.
- Orientation errors for individual seeds on clinical images were also evaluated.
Conclusions:
- The proposed algorithm offers robust and accurate radioactive seed localization in various 3D US images.
- It provides precise seed positioning and orientation, crucial for adaptive radiotherapy.
- The method successfully determines five degrees of freedom for cylindrical seeds.
Keywords:
3D Ultrasound imageBayesian classifieriterative closest point (ICP)prostate brachytherapyradioactive seed localizationsum of squared differences (SSD)support vector machine (SVM)
