Related Experiment Video
Updated: Dec 30, 2025
07:13
Author Spotlight: An Efficient and Robust Software for Automated Fusion of Multiple Preclinical Imaging Modalities
Published on: October 27, 2023
1.6K
Phase mutual information as a similarity measure for registration
Matthew Mellor1, Michael Brady
1Department of Engineering Science, University of Oxford, Parks Road, Oxford OX1 3PJ, UK. matt@robots.ox.ac.uk
Medical Image Analysis
|June 14, 2005
Summary
A novel phase-based method aligns multimodal images by analyzing feature appearance relationships. This approach offers superior robustness to artifacts compared to traditional intensity-based image registration techniques.
Area of Science:
- Medical imaging
- Image processing
- Computer vision
Background:
- Multimodal image registration commonly relies on intensity mapping between images.
- Intensity-based methods struggle with modalities where intensity is not a direct function of tissue class, such as ultrasound.
Purpose of the Study:
- Introduce a new non-rigid alignment method for multimodal images.
- Develop an alternative registration strategy based on local image phase relationships.
Main Methods:
- Model relationships between local image phase, rather than intensity.
- Utilize an image feature-based approach to identify correspondences.
- Compare performance against intensity-based methods.
Main Results:
- The phase-based method enables registration of challenging multimodal image pairs (e.g., ultrasound).
- Performance is comparable to intensity methods under ideal conditions.
- Demonstrates significantly improved robustness to image artifacts.
Conclusions:
- Local image phase modeling provides a powerful alternative for multimodal image registration.
- This feature-based approach enhances registration accuracy and reliability, especially in the presence of artifacts.
