2013 Summer Project Week: Is Neurosurgical Rigid Registration Really Rigid?
- University of Western Australia: Athena Lyons, Karol Miller, Adam Wittek, Grand Joldes
Rigid registration is performed prior to the application of the ISML’s computational biomechanical models of the brain. For application of computational biomechanics it is required that the rigid components of the image (i.e. the skull) be rigidly aligned without scaling. This project investigates the suitability of Slicer's rigid registration method and if, or at what point, brain shift or tumour size effect registration to the extent that image pre-processing is required.
Our approach is to compare the effects of rigid registration on whole images to images where only rigid components are considered for both synthetic samples and hospital data. Analysis will be made using an in house developed Canny Edge detection algorithm. We intend to take advantage of available Slicer experts to better understand the capabilities and limitations of current rigid registration methods and how they may be applicable.
We have improved our understanding of rigid registration in Slicer and determined how to validate the resulting transform file. We also found that resampling in Slicer requires a template/field of view/grid to map the image to that is in the desired sampling dimensions. Most groups use an atlas for this purpose. We have written a loadable extension to generate a grid so that the atlas is not required. For our purposes it generates an isotropic grid and then requests that either linear or lanczos interpolation is used in the resampling. It can be found at . We also learned some methods to produce better masks/label maps.
This work will be delivered to the NA-MIC Kit as a report that can be made available on the NA-MIC web page.
- Intelligent Systems for Medicine Laboratory (ISML) www.mech.uwa.edu.au/ISML
- Mostayed, A., Garlapati, R. R., Joldes, G. R., Wittek, A., Roy, A., Kikinis, R., Warfield, S. K. & Miller, K. [2013. Biomechanical model as a registration tool for image-guided neurosurgery: evaluation against BSpline registration. Annals of Biomedical Engineering, DOI: 10.1007/s10439-013-0838-y.