SAMsrcV3 overview

From MEG Core
Jump to navigation Jump to search

Changes from CTF SAM

Many of the changes made to SAM are intended to simplify the workflow for batch processing – minimizing the user interactions by eliminating copying of required files derived from the MRI, and writing image results to a common named directory. Version 3 analysis programs no longer compute statistical tests for single and multiple comparisons. They compute functions of moment, power, entropy, mutual information, and excess kurtosis. All statistical tests can be done in AFNI (or your favorite neuroimaging software package, but we are partial to AFNI). This provides uniformity and traceability of statistical testing. Instead of .svl files, all image files are written as 3d and 3d+time NIFTI image files for maximum compatibility.

Input and Output File Naming Conventions

All SAMsrc programs make use of a parameter file (ending in .param) that contains all the tunable analysis parameters. For default operation, each function can be run with command line arguments specifying only the parameter file and dataset. The SAMsrc suite also utilizes a command-line parser for flexibility so that the user can utilize data structures not conforming to the default input and output file structures detailed below.

In its default operation, SAMsrc assumes that all MRIs and their derivative files are organized under a single MRI directory root with subdirectories for each subject. The subdirectories are named by character strings derived from the MEG dataset name. The files within each individual's MRI directory are expected to be named according the conventions in the orthohull program.

Intermediate output files (such as covariance and beamformer weights files) are stored in a SAM directory inside the MEG dataset. The output image files for a given analysis are directed to a common directory (designated by the ImageDirectory parameter).

Input MEG Filetypes

This source code, as delivered, is configured to make executables that are compatible with CTF MEG datasets. The software can, however, be configured to process 4D/BTi datasets. In order to compile the software for 4D/BTi compatibility, edit config/Makefile.config to uncomment the following line:


Note that 4D/BTi datasets require two names to uniquely identify which data are to be processed. The dataset name is the directory name containing the data files. This is identified by the -r flag (similar to processing CTF MEG datasets). In addition, the Processed Data File (pdf) is identifed by the -d flag. Both the dataset name and the pdf name are required. Throughout the rest of the documentation, only the usage for CTF MEG datasets will be given.