Creating Personalized Reference Database
As mentioned previously, S²M provides a Reference Database created from T1-weighted MRI data from 188 control subjects, which is freely available for download. This dataset, however, is representative of a Brazilian population, with all images acquired on the same MRI scanner (Philips Achieva, 3T). Although S²M provides different approaches for harmonizing your data with this dataset, creating a study-specific Reference Database may be a more appropriate choice, particularly when differences in population or MRI acquisition parameters are expected and need to be meticulously mitigated.
To create your own Reference Database, you will need:
- T1-weighted images. Recommended: high-quality, isotropic, 3D acquisitions with a voxel size of approximately 1 × 1 × 1 mm³ (±0.1 mm).
- A vector file (.txt, .csv, .xls, or .xlsx) containing the subjects' ages as integer values, with the subjects organized by rows
- A vector file (.txt, .csv, .xls, or .xlsx) containing the subjects' sex, encoded as integer values: 0 = female, 1 = male, with the subjects organized by rows
To start, type the following command in the MATLAB Command Window: SSM_CreateDatabase
A GUI will then open:
Note that you can create a new Reference Database from scratch by either adding raw T1-weighted NIfTI files or adding post-processed files that have already been processed by this S²M tool.
After adding the images, you will need to define the desired smoothing kernel and load the AGE and SEX variable files.
When creating the database from scratch, S²M will generate the gray matter (GM) and white matter (WM) Reference Database files for the selected smoothing kernel, together with formatted files containing TIV, age, and sex information. Note that this process can take several hours. As a rough estimate, allow approximately 15 minutes per image on a good workstation.
If a different smoothing kernel is needed, you can use the post-processed files generated by S²M. In this case, add the post-processed files, select the appropriate map type (GM or WM), define the CAT “Report” folder, select the new smoothing kernel, and load the AGE and SEX variable files.
A separate run is required for each tissue type and each smoothing kernel.
To run the FCD modality, you will need both the GM and WM Reference Database files corresponding to the desired smoothing kernel.
For examples, by running using this setup:
The resultant files in the indicated folder will be:
After creating your own Reference Database, you will need to add it to the S²M Setting:
Run the S²M. in the MATLAB Command Window: s2m or SSM
Click on "Basic Settings" button, and then in "Define Reference Database Folder":
Define the folder we here all your Reference Database files are.
Note that, in the example presented in these instructions, the NEW Reference Database contains only gray matter (GM) parameters, with a smoothing kernel of 8 × 8 × 8 mm³, as defined previously. After running S²M again, the initial checks will display something similar to:
Note the orange warning, indicating that the database currently contains data for only the 8 × 8 × 8 mm³ smoothing kernel and is available only for gray matter (GM) analyses.