AutomaticAnalysisIntroduction - MRC CBU Imaging Wiki

Revision 18 as of 2006-06-02 10:50:26

Clear message
location: AutomaticAnalysisIntroduction

Include(AutomaticAnalysisTopbar)


Contents TableOfContents([maxdepth])


Summary

The system automates the analysis of neuroimaging data using SPM.

Two versions are available: one for Siemens data (e.g., CBU 3T Tim Trio), SPM 5 and NIFTI; and one for Bruker data (e.g., WBIC Medspec 3T), SPM 2 and Analyze. It is written in Matlab.

Main Features

  • Automatic. Almost completely automatic analysis, currently up to the end of normalisation with minimal coding, and the end of second level statistics with a little customisation to specify your first level model.

  • Flexible control. Users that are new to neuroimaging don't get bombarded with options, but are directed to what is essential, while all other settings take sensible defaults. More experienced users can easily change a range of settings, and advanced users can add or replace modules to the system to change any behaviour

  • Restartable. If the users stops the script, or if it crashes (e.g., because of a machine fault, or because the right data aren't there) then when restarted it will begin at the stage where it left off.

  • Automatic good citizen behaviour. Field maps and structurals are automatically detected. Structurals are copied to /imaging/local/structurals/cbu automatically by default.

  • Easy to maintain. The code components are stored in a central repository (/imaging/local/spm/aa) which means that new components can be made available, or existing ones changed in new releases

  • Record keeping. Unlike SPM used from the GUI, the system records exactly which parameters were used, and allows easy recreation of a dataset from the raw data at a later date.

  • Modular, with simple interface. It is easy for Matlab programmers to write new modules and incorporate them into the processing stream.

Getting started

Once installed, to use the Automated Analysis system, first add the paths containing the aa components to Matlab. To do this at the CBU, for Siemens data type

>> aa_ver1

and for Bruker data type

>> aa_ver05

Manuals

For aa version 1.0 and above, this Wiki supecedes the Word versions of the manuals previously provided.

Examples

Example user scripts are in the examples directory of your aa installation. At the CBU, these are

aa version 1.0 (Siemens data)

/imaging/local/spm/aa/aa_ver1.0/examples

aa version 0.5 (Bruker data)

/imaging/local/spm/aa/aa_ver0.5/examples

Contributions

As well as the convenient features, hopefully the main contribution is as a framework we can use to share our analysis knowledge (tricks, code). If you have contributions in the form of: - bug reports - suggestions for improvements in the documentation - new modules - new recipes please send them to me and I will try to fix or incorporate them promptly.

  • Bruker: See how the system works

If you are interested in the guts, the main engine is here: /imaging/local/spm/aa/aa_ver0.4_beta/aa_engine the recipe and parameter set definition files are here: /imaging/local/spm/aa/aa_ver0.4_beta/aa_recipes_and_parametersets the modules are here /imaging/local/spm/aa/aa_ver0.4_beta/aa_modules