As everyone who subscribe to BAMM mailing list know, we are working on a nifti-enabled version of BAMM. Dr John Suckling, Dr Levent Sendur and Dr Alle-Meije Wink and I had been working on it since about July last year. There had been great progress on this front which we will report in detail later.
Today, we are happy to report that a milestone had been reached. The milestone is a simple demonstration data processing using BAMMfx for the fbamm pipeline. This demonstrator is of course available for evaluation on request. Simply send an email to Cinly.
To achieve this milestone, we must first transform the fbamm pipeline to enable it to understand fMRI data supplied in the nifti format then process the data through this pipeline.
This demonstrator have the following shortcomings, most will be rectified in the final product:
- It does not long directory name. This is a direct result of using full path when calling on the command line, i.e., using
/path/to/program /path/to/parameter1 /path/to/parameter2
It is difficult to see how this can be corrected but I will try. - It does not like programs that is verbose in their output. This is a direct result of not emptying the output buffer. Will definitely be corrected in the final product.
- It does not show program output. This is related to (2) and will be resolved.
- It stop halfway through batch processing. This is related to (2). Once the output buffer is full, the program hangs.
- Data in every row must be in the same directory. A few fbamm programs allow specifying only the prefix of the parameters, e.g.,
myprogram /path/to/file
where "file" will be automatically expanded to "/path/to/fileInput1.txt", "/path/to/fileInput2.nii" and "/path/to/fileOutput.nii". However, BAMMfx pipeline language only supports full name in its input, i.e., it is necessary to use "/path/to/fileInput1.txt" and "path/to/fileInput2.nii" instead of "file". To workaround this problem, the prefix "/path/to/file" is derived by finding the common part for "fileInput1.txt" and "fileInput2.nii". If "fileInput1.txt" and "fileInput2.nii" is read from the Input Spreadsheet, then there is a possibility that the inferred prefix is not correct. I did not realized this until I finish testing the demonstrator.
We welcome any suggestions and critcism for this demonstrator.
