11. Exporting and Using C-MIND Data

All data in the database are available for export in compressed NIFTI format (nifti.gz) only. There are two options available for export. a) Raw data and b) Processed data. The data can be exported in bulk as discussed in section 11.2 or using a database query as discussed in section 10.

11.1 Naming conventions for image files

As you export data from the C-MIND database you will discover imaging data in folders for each subject you have included in your database query.  The filenames of these folders and the image data contained within them can be decoded as follows:

Each folder is named as either IRC04H_<xx><F/M><yyy> or UCLA_<xx><F/M><yyy>, where

IRC04H = Children scanned at CCHMC (Cincinnati Children's Hospital Medical Center)

UCLA = Children scanned at UCLA (University of California - Los Angeles)     

xx = Age of the subject at the time of recruitment in years

F/M = Gender (Female/Male)

yyy = Three digit subject ID

 

Raw files:     

Each folder contains list of .nii.gz files.

      <Subj ID>_<P/F>_<1/2/3>_WIP_<Scan name>_<sequence>_1.nii.gz

      eg: Subj_ID = IRC04H_07M003

 

The scans were acquired in 2 sessions.

      P/F = Functional session or Physiological session

 

Children: ages 0-3 and 7-9 scanned at CCHMC were scanned longitudinally for 3 years.

      1/2/3 = represents the year they were scanned

 

Missing files: If any of the raw files are missing, it could be due to 3 reasons:

  • The data was not acquired during the scanning session
  • The data was removed during the initial review process because it failed to meet our quality standards
  • The data was removed after processing because one of the calibration files was not acquired or was insufficient quality.  Calibration files include a B0 field map, a labeling efficiency map, and T1 estimation scans.  If any of these scans were not acquired during an imaging session or were of poor quality, remaining data from the subject could still be included in the database so long as it passed the quality review process. 

Participants were included in the final release database if they had the following imaging data with quality rating <3 (on 0-3 scale).  

Functional: T1, ASL/BOLD functional scan, Resting state BOLD

Physiological: T1 or T2, Baseline CBF, Alpha

 

Scan names:    

Alpha Measurement complex - This is a single-shot gradient echo EPI image acquired in a single slice following a pCASL label which is placed above the carotid bifurcation. The results of this image enter into the formula for quantitative CBF estimation from the ASL data, so an accurate assessment of labeling efficiency is important. The Phase and magnitude images are zipped together to deliver the complex.nii.gz files

Baseline CBF - The cerebral perfusion in the developing brain in children ranging in age from birth to 18 years is a deliverable for this study. This is done non-invasively, without the use of exogenous MR contrast by using a classic pCASL spin labeling pulse MRI method.

T1Est - 100,200,300,400,500,750,1000,1500,2000,3000 - A voxel-wise T1 estimate is also needed in the formula for estimating CBF quantitatively from ASL MRI data.  It is important to estimate T1 on a pixel by pixel basis to get accurate estimates of CBF from the model because T1 varies over the brain and between GM and WM. T1 also varies with age, so it is important to map it in each subject. If these images are not acquired in a subject, an estimated value for T1 for GM and WM are assumed and used in the ASL perfusion estimation pipeline.

T1W - This is a standard Inversion Recovery prepared, T1-weighted, 3D isotropic method is used to acquire a 1x1x1 mm anatomical image for underlayment of functional and DTI maps, GM/WM segmentation, normalization or for VBM or other morphological applications.

T2W - This is a 3D fluid-attenuated, inversion-recovery (FLAIR) sequence with long inversion time to suppress CSF signal and long TE to emphasize T2 contrast between GM and WM.  This is acquired in the same FOV and resolution (1x1x1 mm) as the T1W image and can be used in multi-spectral image segmentation or for VBM or other morphological applications

ASLBOLD  - The ASL/BOLD acquisition is a new method that is designed to allow for simultaneous acquisition of an ASL perfusion scan interleaved with a BOLD functional scan within a few hundred milliseconds.

Stories Task = The Story Processing Task involves passive listening to stories read in a mother's voice using language at the pre-school level, during the “stimulated” condition. The baseline condition is a non-speech broadband noise sweep with center frequencies ranging from 200-4000Hz. The duration of the sweep across this frequency range varies from 0.5-2.0 Hz. This baseline was designed be similar to the frequency range of speech but very non-speech like because it avoids the rapid temporal transitions involved in speech. The task begins with a stories block (64 seconds) followed by a noise block (64 seconds). A total of 5 cycles through the two conditions results in a total scan time of 10 minutes 40 seconds.  Concurrent ASL/BOLD fMRI data is aquired during this language stimulation paradigm.

Sentences Task = In the Sentence-Picture matching Task, the “stimulated” condition involves auditory presentation of a simple active or passive sentence simultaneously with visual presentation of a picture. The “baseline” condition involves single nouns also presented auditorily, repeated four times so that the noun stimuli are similar in duration to sentence stimuli. In both conditions, the participant’s task is to make a button-press response indicating if the auditory stimulus accurately describes the picture. Each block includes 15 trials (4250 ms each), and these blocks are counter-balanced for yes/no responses, and sentence type. The task begins with a nouns block followed by a sentences block. A total of 5 cycles through the two conditions results in a total scan time of 10 minutes 40 seconds. Concurrent ASL/BOLD fMRI data is aquired during this language stimulation paradigm.     

Resting State - This is a typical BOLD-weighted EPI gradient echo scan acquired in the resting state. No task behavior is required and the subject is either asleep or awake with visual fixation on a cross hair.

GRE Map abs and rads – It is the phase difference in radians of delta TE at 2.3 msec. The phase encoding direction is AP. It is used for the field map correction.

 

Processed files:

     

Each folder contains 'F' or 'P' or both sub-folders. In each folder, the following image files are present.

 

The 'F' sub-folder contains:

<SubjID>_F_<1/2/3>_ASL_Stories_Level_1.tar.gz, <SubjID>_F_<1/2/3>_BOLD_Stories_Level_1.tar.gz

<SubjID>_F_<1/2/3>_ASL_Sentences_Level_1.tar.gz, <SubjID>_F_<1/2/3>_BOLD_Sentences_Level_1.tar.gz          

Note that Sentence Task was only performed in subjects age >=7 years. Stories Task was performed in all subjects.      

Outputs of the first level fMRI analysis are stored in a .tar.gz file containing the full first level FEAT folder from FSL. After extracting these folders for a set of subjects they are ready for use in a second level analysis. Registration to standard space has already been performed via ANTs SyN. No registration should be run at the second level.

      The structure of the various files and subfolders present are described in detail at the FSL wiki: FEAT Output Details

For ASL (either paradigm):       

            Contrast 1: positive BOLD activation

            Contrast 2: positive ASL activation

            Contrast 3: positive ASL baseline

            Contrast 4: negative BOLD activation (deactivation)

            Contrast 5: negative ASL activation (deactivation)  

For BOLD (either paradigm): (Stats folder)           

            Contrast 1: positive BOLD activation

            Contrast 2: negative BOLD activation (deactivation)

The above listed contrasts are located in 'FEAT_Process_<1/2>.Output_dir-1/Stats' folder.

<SubjID>_F_<1/2/3>_T1Segment_<CSF/GM/WM>.nii.gz -  partial volume estimate from FAST-based segmentation for the Grey Matter/White Matter/Cerebrospinal Fluid.

<SubjID>_F_<1/2/3>_T1W_Structural_N4_brain.nii.gz - T1 anatomical after intensity normalization and brain extraction.

<SubjID>_F_<1/2/3>_wlin_T1Segment_<CSF/GM/WM>.nii.gz - Grey Matter (GM)/ White Matter (WM)/ Cerebrospinal Fluid (CSF) partial volume estimate, affine transformed to MNI standard space.

<SubjID>_F_<1/2/3>_wANTS_T1Segment_<CSF/GM/WM>.nii.gz - Grey Matter (GM)/ White Matter (WM)/ Cerebrospinal Fluid (CSF) partial volume estimate, after nonlinear deformation transformed to MNI standard space.

<SubjID>_F_<1/2/3>_T1_to_MNI_lin.nii.gz - Processed Structural Image, affine transformed to MNI standard space (using FSL FLIRT).

<SubjID>_F_<1/2/3>_T1_to_MNI_ANTS_SyN_Deformed.nii.gz - Processed Structural Image, after nonlinear deformation to MNI standard space using (ANTS Syn).

 

The 'P' sub-folder contains:

<SubjID>_P_<1/2/3>_myalpha.txt - This is a number corresponding to the estimated pCASL inversion efficiency.

<SubjID>_P_<1/2/3>_CBF_Wang2002.nii.gz - This is the estimated quantitative CBF map, in the low-resolution subject space.

<SubjID>_P_<1/2/3>_wlin_CBF_Wang2002.nii.gz - BaselineCBF Map, affine transformed to MNI standard space (using FSL FLIRT).

<SubjID>_P_<1/2/3>_wANTS_CBF_Wang2002.nii.gz - BaselineCBF Map, after nonlinear deformation to MNI standard space (using ANTS syN).

<SubjID>_P_<1/2/3>_T1_map_reg2CBF.nii.gz - estimated T1 relaxation time map, registered to CBF, in the low-resolution subject space.

<SubjID>_P_<1/2/3>_T1Segment_<CSF/GM/WM>.nii.gz -  partial volume estimate from FAST-based segmentation for the Grey Matter/White Matter/Cerebrospinal Fluid.

<SubjID>_P_<1/2/3>_T1W_Structural_N4_brain.nii.gz - Brain Extracted Structural - T1 anatomical after intensity normalization and brain extraction.

<SubjID>_P_<1/2/3>_wlin_T1Segment_<CSF/GM/WM>.nii.gz -  Grey Matter (GM)/ White Matter (WM)/ Cerebrospinal Fluid (CSF) partial volume estimate, affine transformed to MNI standard space.

<SubjID>_P_<1/2/3>_wANTS_T1Segment_<CSF/GM/WM>.nii.gz - Grey Matter (GM)/ White Matter (WM)/ Cerebrospinal Fluid (CSF) partial volume estimate, after nonlinear deformation transformed to MNI standard space.

<SubjID>_P_<1/2/3>_T1_to_MNI_lin.nii.gz - Linear (FLIRT) Normalized T1, affine transformed to MNI standard space (using FSL FLIRT).

<SubjID>_P_<1/2/3>_T1_to_MNI_ANTS_SyN_Deformed.nii.gz - Nonlinear (ANTS) Normalized T1 - after nonlinear deformation to MNI standard space using (ANTS Syn).

 

11.2 Bulk Exporting Imaging Data

 

Click on “Query and Processing” -> “Download all imaging data” in the database interface to get a collection of tar files as in the following Screenshot (Figure 11.1).

Figure 11.1: Bulk download of image files through the database web interface.

 

11.3 Bulk Exporting Behavioral Data

Currently the only way to download behavioral data is through the database query interface. For details please see Section 10 of this manual. To do this you must register for the database.  This is a requirement of NICHD for management and tracking of the C-MIND data.

11.4 Submitting the data to a LONI pipeline for processing

This feature is currently only available to CCHMC users. We will shortly be making an option to execute LONI pipeline workflows on C-MIND data on the USC LONI cluster available to the general public!

In the meantime, to process C-MIND data using the pipelines we have constructed, you may download the NiPype pipelines from bitbucket and run them locally on your desktop or other computing resources.  For large data sets (more than 20 subjects) we recommend using a computing cluster or planning on having a cup of coffee or two if you run the job on your desktop.