jist:MultiCRUISE

From NITRC Wiki

Contents 1 Multi-modal CRUISE 1.1 Summary of Functionality 2 Processing Steps 2.1 Specifying modality inputs 2.2 Co-Registration 2.3 CATNAP DTI Processing 2.4 CRUISE Cortical Surface Reconstruction 2.5 Lesion Segmentation with Lesion TOADS 2.6 Multi-Atlas Gyral Labeling 2.7 Connectivity Matrices 2.8 Running everything

Multi-modal CRUISE

Multi-modal CRUISE encompasses the functionality of a number of other JIST layouts, namely the CRUISE and CATNAP layouts which perform cortical reconstruction and end-to-end DTI processing, respectively. Furthermore provides processing using modalities in a complementary fashion, most notably the computation of connectivity matrices between cortical gyri. The cortical gyri are computed as part of the CRUISE pipeline, while metrics between regions can be computed over volumes connecting ROIs as defined by DTI fiber tractography. It also enables the co-registration of all modalities. The layouts here enable the registration of Pib PET and CBF PET to a structural MR (T1) acquisition.

Summary of Functionality

1. Cortical Reconstruction using the CRUISE pipeline.
2. DTI processing and Fiber tracking using the CATNAP pipeline.
3. Lesion segmentation using Lesion TOADS.
4. Multi-atlas gyral labeling.
5. Connectivity Matrix computation.
6. Coregistration of MR modalities (MPRAGE, SPGR, FLAIR, T1w, T2w)
7. Coregistration of PET modalities to structural MRI. (PiB PET, CBF (Cerebral blood flow) PET)

Processing Steps

• Download the MultiModal CRUISE layout: http://www.nitrc.org/frs/?group_id=228
• Sample data is not available at this time due to data-use restrictions.
• Load the layout file.

This layout contains a number of sub-processing pipelines, some of which are duplicates of single-modality processing. These sub-layouts are highlighted below.

Specifying modality inputs

• Each blue input source corresponds to a modality. Files for all modalities to be processed must be specified.
• For each modality, create a text file that lists all files to be processed with each subject on one line.

• • For example, a text list for MPRAGE files for three subjects would look like this:

/home/bennett/data/subject1_mprage.rec
/home/bennett/data/subject2_mprage.rec
/home/bennett/data/subject3_mprage.rec

• Note: repeated scans in DTI must be specified together. Details can be found on the CATNAP page.
• Next, for each modality, select each blue input source and select the text file that was created corresponding to that modality.

Co-Registration

The top right sub-processing pipeline shown at right enables the co-registration of all modalities of interest. This is accomplished by using the Optimized Automatic Registration algorithm in MIPAV. Select "MPRAGE" and specify a text file containing MPRAGE files which will serve as the registration target. Next, specify input files for all the other modalities of interest. Here are shown:

1. MPRAGE
2. T1w
3. T2w
4. SPGR
5. FLAIR
6. meanPIB
7. paraPIB

CATNAP DTI Processing

The top-left sub-processing pipeline duplicates the CATNAP pipeline with Par v4 Files and eddy current geometry correction. Information for running this sub-pipeline can be found on the CATNAP page.

CRUISE Cortical Surface Reconstruction

The top middle sub-processing pipeline duplicates the CRUISE cortical surface reconstruction pipeline. Information for running this sub-pipeline can be found on the CRUISE page.

Lesion Segmentation with Lesion TOADS

The bottom right processing pipeling shown at right segments White-Matter lesions using Lesion TOADS. It works best with at least MPRAGE and FLAIR images, though T2, PD, and SPGR and be included in the segmentation algorithm as well. In the pipeline shown, LesionTOADS will automatically be run if MPRAGE, SPGR, FLAIR, and T2 acquisitions are specified. Additional modalities can be included by replicating the "Mask" module and connecting it's output to LesionTOADS. Modalities can be excluded by disconnecting their inputs to LesionTOADS.

Multi-Atlas Gyral Labeling

The bottom left portion of the multi-modal CRUISE layout contains modules that automatically generates gyral labels on the cortical surfaces created by CRUISE. This is performed by first non-linearly registering a collection of volumetric atlases to the target. Then, previously segmented gyral labels on the atlases are transferred to the target volume using the mapping from the registration. Finally, the transferred segmentations are mapped to the CRUISE surface.

• To perform the labeling, the locations of the atlases and atlas labels for the "Labeling via Multiple Atlas Registration" module must be specified. Atlases will soon be available for download here.
• The pipeline will automatically label the CRUISE surface associated with the target.

Connectivity Matrices

The bottom left portion of the multi-modal CRUISE layout also contains modules to compute the connectivity matrices between the labeled ROIs. Since the computation of connectivity matrices requires that both CATNAP (CATNAP instructions) and CRUISE (CRUISE instructions) be run, the inputs for each of these sub-pipelines must be specified.

• When the pipeline is run, the connectivity matrices will automatically be computed using the outputs of CATNAP, CRUISE, and Gyral Labeling.

An example cortical surface with gyral labels

An example connectivity matrix

Running everything

• Open the JIST process manager.
• Select Project->Layout Preferences (1) and select an output directory (2). Optionally, adjust the maximum amount of memory or maximum number of simultaneous cores that will be run.

• Verify that all modules are either "READY" or "NOT READY". If the status shows "FAILED" or "OUT-OF-SYNC", then you have selected an output directory that already contains processed results. You may remove these existing results at your own risk with Scheduler->Clean All.
• When ready, click Scheduler->Start Scheduler.
  • For more information on the process manager, see jist:ProcessManager
  • For more information on the layout tool, see jist:LayoutTool
• You should now observe that experiment status change to "RUNNING" and eventually "COMPLETED"