Related Experiment Videos
Automatic MR volume registration and its evaluation for the pelvis and prostate
Baowei Fei1, Andrew Wheaton, Zhenghong Lee
1Department of Biomedical Engineering, Case Western Reserve University, Cleveland, OH 44106, USA.
Physics in Medicine and Biology
|April 5, 2002
Summary
A new 3D mutual information registration method accurately aligns prostate MRI scans. This robust technique improves prostate cancer diagnosis and treatment planning by minimizing registration errors.
Area of Science:
- Medical Imaging
- Image Registration
- Computational Anatomy
Background:
- Accurate registration of pelvic MRI is crucial for prostate cancer management.
- Existing registration methods may lack robustness in clinical settings.
Purpose of the Study:
- To develop and evaluate a robust 3D mutual information registration method for prostate MRI.
- To assess registration accuracy under various clinical conditions for prostate cancer applications.
Main Methods:
- Implemented a multi-resolution approach using correlation coefficient and mutual information as similarity metrics.
- Incorporated a restart strategy with random parameter perturbation to avoid local minima.
- Tested the method on MRI volumes acquired in diagnostic and treatment positions, with varying bladder filling and time intervals.
Main Results:
- Achieved high accuracy in registration, with bony landmarks and prostate centroids aligned within approximately 1.6 mm and 1.4 mm, respectively, under comparable conditions.
- Observed a posterior shift of 2.8-3.4 mm in prostate centroids between diagnostic and treatment positions.
- Identified the need for manual cropping to register images from different positions.
Conclusions:
- Automatic rigid body registration demonstrates sufficient accuracy for many prostate cancer applications.
- Minimizing prostate displacement for interventional MRI-guided treatments can be achieved by standardizing imaging positions and reducing bladder/rectal volumes.