This algorithm takes two MRI models of different contrast as input, matches the contrast of one model to that of the other, and returns a contrast-matched model. The algorithm thereby enables non-linear coregistration using cross correlation of multi-modal minimum deformation averaged MRI models.
For the algorithm to be successful, the two models must initially be roughly aligned.
The algorithm contains the following steps:
- The intensity range of both models is normalised to be between 0-100
- A mask, generated using BET2, is applied to both models
- The model with the contrast, to which the other model is matched to, is blurred
- Both models are preprocessed in preparation for the lookup
- A voxel intensity value lookup between the two models, based on voxel location and majority decision, is performed
- A spline function, that describes the voxel intensity relation between the two models, is determined
- A lookup table is generated on the basis of the spline function
- One of the models is converted using minclookup and the generated lookup table
The algorithm is implemented as a script in Jupyter Notebook.
The script uses external software, so for it to be executable, the following software is required installed:
- Brain Extraction Tool (BET2). Can be found here: https://github.com/liangfu/bet2
- MINC toolkit version 1.9.11. Can be found here: https://bic-mni.github.io/#v2-version-19111911
- Bhimal Ramsing
- Henar Rituerto
- Julie Broni Munk
- Nina Jacobsen
- Maciej Plocharski
- Lasse Riis Østergaard
- Lars Marstaller
- David Reutens
- Markus Barth
- Andrew Janke
- Aswin Narayanan
- Steffen Bollmann