Related Experiment Videos
Robust nonrigid registration to capture brain shift from intraoperative MRI
Olivier Clatz1, Hervé Delingette, Ion-Florin Talos
1Epidaure Research Project, INRIA Sophia Antipolis, 06902 Sophia Antipolis Cedex, France. olivier.clatz@sophia.inria.fr
IEEE Transactions on Medical Imaging
|November 11, 2005
Summary
This study introduces a novel algorithm for registering 3-D brain magnetic resonance (MR) images, effectively correcting for brain shift during surgery. The method ensures accurate intraoperative guidance by robustly estimating tissue deformation.
Area of Science:
- Medical imaging
- Neurosurgery
- Computational anatomy
Background:
- Intraoperative brain shift complicates accurate surgical navigation.
- Preoperative magnetic resonance (MR) images require registration with intraoperative images for guidance.
Purpose of the Study:
- To develop and validate a robust algorithm for 3-D intraoperative brain MR image registration.
- To address the challenge of brain shift by accurately estimating and correcting tissue deformation.
Main Methods:
- A novel iterative registration framework combining approximation and interpolation formulations.
- Robust estimation of displacement fields using sparse, noisy data and outlier rejection (weighted least trimmed squares).
- Patient-specific finite element models for realistic brain tissue mechanics and parallelized computation for clinical time constraints (35s registration).
Main Results:
- The algorithm successfully registered 3-D brain MR images with significant brain shift (up to 14 mm) in six brain tumor resection cases.
- Demonstrated good recovery of large displacements.
- Showed limited accuracy decrease near the tumor resection cavity.
Conclusions:
- The proposed algorithm offers a robust and efficient solution for intraoperative brain MR image registration.
- It effectively compensates for brain shift, enhancing surgical accuracy and navigation.
- The method's speed and accuracy make it suitable for clinical application.