Related Experiment Video
Updated: Jul 7, 2025
02:09
Multi-modal Pulmonary Imaging: Using Complementary Information from CT and Hyperpolarized 129Xe MRI to Evaluate Lung Structure-Function
Published on: April 12, 2024
613
Comparison of physics-based deformable registration methods for image-guided neurosurgery.
Nikos Chrisochoides1, Yixun Liu1, Fotis Drakopoulos1
1Center for Real-Time Computing, Computer Science Department, Old Dominion University, Norfolk, VA, United States.
Frontiers in Digital Health
|December 25, 2023
Summary
This study enhances non-rigid registration accuracy by improving outlier rejection methods, reducing errors by 2.5mm. It also explores Quantum Computing for faster, more robust medical image registration.
Area of Science:
- Medical image analysis
- Computational anatomy
- Scientific computing
Background:
- Physics-based non-rigid registration is crucial for medical imaging.
- Brain tumor removal causes significant shifts, challenging registration accuracy.
- Existing methods struggle with outliers from sparse, noisy, or incomplete data.
Purpose of the Study:
- To compare finite element-based methods for non-rigid registration.
- To improve registration accuracy in the presence of brain shift outliers.
- To investigate the potential of Quantum Computing for registration acceleration.
Main Methods:
- Comparison of three finite element-based non-rigid registration techniques.
- Implementation of approximation- and geometry-based outlier rejection.
- Evaluation of registration error reduction and real-time performance.
Main Results:
- Outlier rejection improved rigid registration error by 2.5mm.
- The enhanced method achieved real-time constraints within 4 minutes.
- Preliminary results show Quantum Computing's potential for accelerating feature detection and block matching.
Conclusions:
- Combined outlier rejection significantly enhances registration robustness.
- Open problems remain for improving registration with sparse, noisy data.
- Quantum Computing offers a promising avenue for computationally intensive registration tasks.
