Related Experiment Video
Updated: Jun 28, 2026
12:09
Patient-specific Modeling of the Heart: Estimation of Ventricular Fiber Orientations
Published on: January 8, 2013
An atlas-based segmentation propagation framework locally affine registration--application to automatic whole heart
Xiahai Zhuang1, Kawal Rhode, Simon Arridge
1Centre for Medical Image Computing, Med Phys Dept, UCL, WC 1E 6BT, UK. x.zhuang@ucl.ac.uk
Medical Image Computing and Computer-Assisted Intervention : MICCAI ... International Conference on Medical Image Computing and Computer-Assisted Intervention
|November 6, 2008
Summary
This study introduces a new registration algorithm for medical imaging, preserving anatomical details for accurate whole heart segmentation. The method enhances segmentation accuracy in cardiac MRI scans.
Area of Science:
- Medical Image Analysis
- Computational Anatomy
- Biomedical Engineering
Background:
- Accurate segmentation of anatomical structures from medical images is crucial for diagnosis and treatment planning.
- Existing registration algorithms may struggle to preserve local anatomical and intensity relationships, impacting segmentation accuracy.
- Cardiac magnetic resonance imaging (MRI) presents challenges for automated segmentation due to complex anatomy and motion.
Purpose of the Study:
- To develop and validate a novel registration algorithm for accurate whole heart segmentation from cardiac MRI.
- To preserve local anatomical and intensity class relationships during registration.
- To improve the robustness and accuracy of atlas-based segmentation propagation frameworks.
Main Methods:
- A novel registration algorithm for locally affine transformations was developed.
- The method incorporates a regularization procedure for global diffeomorphic transformation.
- A generic method for accurate inversion of the deformation field was implemented within an atlas-based segmentation propagation framework.
- The algorithm was applied to automatically segment the whole heart in 18 cardiac MRI volunteers.
Main Results:
- The proposed method provided robust initialization for atlas-based segmentation propagation.
- The registration algorithm successfully preserved local anatomical and intensity relationships.
- Validation against other registration strategies demonstrated improved accuracy in whole heart segmentation.
- Achieved an average segmentation accuracy of 1.8 +/- 0.42 mm.
Conclusions:
- The novel registration algorithm offers a robust and accurate solution for whole heart segmentation in cardiac MRI.
- Preservation of local anatomical and intensity relationships is key to improving segmentation accuracy.
- The method enhances atlas-based segmentation propagation frameworks, demonstrating significant potential in clinical applications.
