Registration
use –action=
regto apply linear and/or nonlinear registration.
Update: (2025/03/03) The registration syntax has revised.
Examples
*Apply linear and nonlinear registration to nonlinearly warp subject’s QA and ISO maps to template QA and ISO maps (duel modality warping)
dsi_studio --action=reg --source=subject_qa.nii.gz,=subject_iso.nii.gz --to=template_qa.nii.gz,template_iso.nii.gz --output_mapping=mapping_field.mz
*Warp the subject image to the template space using previously computed mapping field
dsi_studio --action=reg --source=subject_qa.nii.gz --mapping=mapping_field.mz
*Unwarp the template image to the subject space using previously computed mapping field
dsi_studio --action=reg --to=template_qa.nii.gz --mapping=mapping_field.mz
*Warp all nifti image and all tract files from the template space to the subject space
dsi_studio --action=reg --source=*.nii.gz --mapping=mapping_field.mz
Registration Functions
The function warps the subject/source image to the template/target image
| Parameters | Description |
|---|---|
| source | specify the nifti file(s) of the subject/source image. You may specify multiple modalities separated by comma |
| to | specify the nifti file of the template/target image. You may specify multiple modalities separated by comma |
| output_mapping | (optional) specify the file to store the mapping field (e.g., –output_mapping=mapping.mz) |
| mapping | specify the previous computed mapping and applied to –source or –to , which can take multiple nii.gz or tt.gz files separated by comma. |
| Parameters | Default | Description |
|---|---|---|
| cost_function | mi | specify the cost function to linear registration. mi: mutual information cc: correlation coefficient |
| reg_type | 1 | specify whether nonlinear registration is needed. set it to 0 if only linear registration is needed |
| normalize_signal | 1 | specify whether pre-registration signal normalization is needed. The normalization scales the image value between 0 and 1 |
| resolution | 2 | Specify the final resolution. The value can be 1,2,4,8. 2 means the nonlinear registration will be conducted at x2 down sampling. |
| speed | 1 | specify the convergence speed for nonlinear registration |
| smoothing | 0.2 | specify the mapping field smoothing for nonlinear registration. The value can be 0 (no smoothing) or 0.95 (more smoothing applied) |
| iterations | 200 | specify number of iterations applied to compute nonlinear registration |
| min_dimension | 8 | specify the minimum dimension to start the nonlinear registration. |