This tool converts the structure of the input DICOM data to align with the BIDS convention and allows to plug the data into BIDS-tools.
Neuroimaging experiments result in complicated data that can be arranged in many different ways. So far there is no consensus how to organize and share data obtained in neuroimaging experiments. Even two researchers working in the same lab can opt to arrange their data in a different way. Lack of consensus (or a standard) leads to misunderstandings and time wasted on rearranging data or rewriting scripts expecting certain structure. The Brain Imaging Data Structure (BIDS) describes a simple and easy to adopt way of organizing neuroimaging and behavioral data. More info here.
The DICOM folders are processed used dcm2niix. The output is a NIfTI and a JSON files with the information required. The files are renamed following the convention and stored in the corresponding folders. Please check the BIDS structure documentation for more information.
The output container content depends on what the input had. For a simple example, if we had a T1 DICOM file, the output would appear like this.
- sub-01/anat/sub-01_T1w.nii.gz
- sub-01/anat/sub-01_T1w.json
The label in this case is '01', identifies the subject. The structural images are stored in the anat/ folder. The functional images are stored in the func/ folder. The diffusion images are stored in the dwi/ folder. The PET images are stored in the pet/ folder. Depending on the acquisition parameters, the tool might add other entities in the file name. The specifications for the MRI data can be found in the BIDS structure website. The tool further checks the REQUIRED parameters are included in the JSONs.
Validation can be done with:
The latter is implemented within the tool by its end. It modifies the QA status if PASS/FAIL.
The tool is adapted for (Brain Imaging Data Structure v1.8.0):
- Structural MR images:
- T1w
- T2w
- PDw
- T2starw
- FLAIR
- PDT2
- UNIT1
- Homogeneous (flat) T1-weighted MP2RAGE image
- MRA
- fMRI
- Diffusion
- Fieldmaps
- Case 1 (Phase-difference map) (standard for fMRI)
- Case 4 (Multiple phase encoded directions ("pepolar")) (standard for DWI, if for fMRI acquire several volumes [7,12])
- MoCo files are not considered for this case.
- PET (Considerably depends on the input settings.)
For the fieldmap usage multiple combinations of them could be provided to a tool, and also different set of tags within the json. Thus, there are certain rules to establish priorities in the different settings and the different tags. Next we list those priorities of nipreps:
- Use of B0FieldSource
- Direct fieldmaps (_fieldmap.nii) specified with IntendedFor
- Phase-difference fieldmaps (_phasediff.nii) specified with IntendedFor
- Two-phase fieldmaps (_phase{1,2}.nii) specified with IntendedFor
- PEPOLAR fieldmaps (2x _epi.nii) specified with IntendedFor
- Reverse-phase-encoding fieldmaps (*_epi.nii where the alternative direction is _bold.nii) specified with IntendedFor
Entities supported (Brain Imaging Data Structure v1.8.0):
- sub
- task
- acq
- ce
- rec
- dir
- run
- echo
- trc
However, the tool is not adapted for (Brain Imaging Data Structure v1.8.0):
- ASL
- inplaneT1?
- T1 weighted structural image matched to a functional (task) image.
- inplaneT2?
- T2 weighted structural image matched to a functional (task) image.
- MRS
- EEG
- B0 mapping
- Derivatives
- CBV (Cerebral Blood Volume contrast)
- sbref (Single-band reference for one or more multi-band dwi images.)
- Blood recording data (PET)
Parametric maps (T1map, R1map, T2map, R2map, T2starmap, R2starmap, PDmap, MTRmap, MTsat, T1rho, MWFmap, MTVmap, PDT2map, Chimap, TB1map, RB1map, S0map, M0map) - Entities (ses, flip, inv, mod, sample, stain, mt, part, proc, hemi, space, split, recording, chunk, res, den, label, desc)
As for the common metadata fields shown next (Brain Imaging Data Structure v1.8.0). Most of them are just 'recommended' being only available when extracted by the converter.json. If they are obligatory for a given modality, the file information extracted by the upload tool is further checked.
- Scanner Hardware
- Sequence Specifics
- In-Plane Spatial Encoding
- Timing Parameters
- RF & Contrast
- Slice Acceleration
- Anatomical landmarks
- Echo-Planar Imaging and B0 mapping
- Institution information
Current limitations:
- The tool supports multiple runs and multiple echoes. However, it does not support the two of them simultaneously (requires a lot of dev).
- In PET when the same acq. is repeated with different reconstruction it is considered as different runs and does not follow the standard to add the counter to the rec entity.
Multiple acqs with different TR will remove the latter input to the tool. Example: hrad_3 for DWI with TR: 2823 and 2832. Same happens with other parameters such as TE, Voxel Size, ImageType. Enabling this would considerably complicate file names since the only possible differences included in entity: acq. - 4D MoCo are output as they are initially converted according to the initial information we have on them. 3D MoCo are removed. They are not use neither as target nor as source of fmaps.
- Nifti support is not supported! Since it is considered useless.
- All of the metadata that is possible to be extracted from the DICOM header will be extracted from there. But not all is included. By default, I've included in the settings those that I've never seen, but depending on the data itself we might have to add more metadata in the settings.
Create free account now!
