- A set of brains in a longitudinal study, in either native (scanner) space or after a standard lsq6 alignment. Currently, we recommend lsq6. (See note below.)
- A csv file containing all brains in the study, with the following format:
- Each row in the csv file will correspond to one subject
- Each row will contain a comma separated list of scans for that subject, with the earliest time point first, followed by the other time points in chronological order.
- Absolute path names should be used for all file names.
- Optional: A directory of masks, which can contain either one mask for all brains, or one uniquely identified mask for each brain.
- Optional: A completed registration (iterative model building, a.k.a MICe-build model) for all subjects at a common time point. (e.g. In the example above, this would be all P42 scans). If this has not been done prior to running the registration chain, it can be done as part of the registration_chain.py call. (See note below.)
|title||Important Pre-requisite Caveats|
Pre-requisite #1: If input scans are in native space, they will be aligned first using the LSQ6 module, prior to starting the chain alignment.This happens regardless of whether
--no-run-groupwise is specified.
- . Table must have atleast three columns: subject_id, timepoint, and filename. 'filename' is the MR images and can be either relative or full paths.
- A csv file with init models. It has two columns: time_point and modelfile. This way you can have different init models at different timepoints in the registration.
Pre-requisite #1: If input scans are in native space, they will be aligned first using the LSQ6 module, prior to starting the chain alignment.
The latest tagged version of this code (available on github) is installed at MICe. To run at MICe, simply type registration_chain.py with the appropriate options, or --help for a complete list. To install and run this elsewhere, see the pydpiper wiki page and github repository for more information.