Posted in:

Description

The HOF pipline implements an online interface to Multimodal Glioma Analysis (MGA), a preprocessing suite for multi-modal tumor MR imaging for ROI analysis.

Specific actions performed:
1. Auto-selects best available T1 weighted image as session registration target;
2. Auto-selects best available target T1 weighted brain atlas;
3. Generates DWI and DSC parametric maps;
4. Co-registers all structural and physiological images to the T1w target;
5. Resamples all images in 4dfp format to target and to (1,1,1) atlas space;
6. Generates registration QC report;
7. Uploads resampled coregistered images to HOF_reg and HOF_QC resources of the session.

Can be run on

HOF.xml can be used on MR Sessions with at least one structural scans. DTI and DSC sequences will be coregistered to structural T1 weighted target.

Project Inputs

Name
Description
Allowed Values
Launch Automatically
Indicate if the pipeline should be run on every
scan as soon as it is archived
unchecked
Scans
ID (scan number) for the scans to be included
text
HOF IDs
HOF ID (scan type label) for those scans
text
Siemens scanner
Indicate if a Siemens scanner was used
0,1
0 = no
1 = yes
Manual Params
Specify if automatic parameters should be overwritten
strings*
none = do not override
CONDR script path
The physical location of the MGA scripts
directory path

Launch Inputs

Name
Description
Allowed Values
Mosaic [Siemens] DTI encoding
Expect mosaic DTI encoding
checked, unchecked
Override automatic parameters
Specify if automatic parameters should be overwritten
strings*
none, blank = do not override
Scan
Indicate which of the listed scans should be run
checked, unchecked
HOF ID
HOF IDs for those scans
select or leave as preselected
Send additional confirmation emails
Enter the email address for other users who should
be notified upon successful completion of scan
email addresses separated by commas
*Parameters are entered in the form <name>=value, separated by comma or newline. Currently supported manual parameters with usage examples:
  • dwi_nx=64,dwi_ny=64 diffusion slice dimensions in mosaic
  • targ_t1=4 set T1 weighted registration target to scan 4
  • targ_t1=S001_Preop:4:MPRAGE set T1w registration target to scan 4 from session S001_preop, with HOF_Id=MPRAGE.
  • targ_t2=5 Set T2 weighted image to use for DSC and DTI processing to scan 5
  • targ_t2=S001_preop:5:T2lo Set T2 weghted image for DSC and DTI processing to scan 5 from session S001_preop, with HOF_Id=T1lo.
  • dwi1=5,dwi2=6 scan IDs for split DTI series (local series only)
  • dwi_zflip=1 z flip DTI scan
  • perf_zflip=1 z flip perfusion DSC scan
  • atlas=CAPIIO specify target brain atlas. Possible values:
    711-2A 711-2B 711-2C 711-2F 711-2K 711-2L 711-2O 711-2O 711-2S 711-2V 711-2Y 789yo_atlas_111 ALLEGRA_O_111 ALLEGRA_Y_111 ALLEGRA_YO_111 CAP_CKY_111 CAPIIO GENPIBO LSIIC LSIIK LSIIO LSIIY MSADEM_K_111 MSADEM_Y_DC newKIDADULT_ATLS_111 NP720_M_DC NP720_M_NDC NP891_YO_NDC_111 P44O_C_111 P44W_C_111 PDSIIBRAIN_111 small_WM stealth_atlas_11to18 stealth_atlas_3to10 TRIO_5yo_NDC TRIO_KY_NDC TRIO_Y_NDC
  • custom_dti=1 auto-choose DTI direction file from NIL dir
  • dti_prm=25dir_b1000.prm use specific DTI .prm file from NIL dir.
    possible options are:12dir_x2_b1400.prm DiffVec_1000_25dirplusB0.prm MDDW_12dir_3b_dti.prm
    12dir_x3_b1000.prm DiffVec_1400_23dirnoB0.prm phillips_hex_b1000.prm
    12dir_x4_b1000.prm DiffVec_1400_25dirnoB0.prm Siemens_30dir_b1000.prm
    17+16+15_b1200_NOreorder.prm DiffVec_1400_25dirplusB0.prm symphony_12_b1000.prm
    20dir_b1000.prm GE_25dir_b1000.prm symphony_12_b14000.prm
    20dir_x4_b1000.prm GE_26dir_b1000.prm symphony_12_b800.prm
    DiffVec_1000_25dirnoB0.prm HARD12x4_b800.prm symphony_hex_b1000.prm
    25dir_b1000_GE_dcm.prm 26dir_b1000_GE_dcm.prm

Additionally, you can list scans other than targ_t1 and targ_t2 to import from other sessions of this subject, separated by comma or newline.
Format: <HOF Id>=<session label>:<scan id>, e.g.
T1hi=S001_Preop:4,T1hi=S001_Postop:5,T2FLAIR=S001_Preop:7
Some commonly imported scan types:
  • MPRAGE T1 weighted, ~1mm isotropic voxels, MPRAGE or similar contrast
  • T1hi T1 weighted, high in and out of plane resolution, e.g. 1x1x2.5mm voxels
  • T1lo T1 weighted, lower in and/or out of plane resolution, e.g. 2x2x3mm voxels.
  • T2FLAIR T2 weighted, FLAIR or similar contrast.
  • T2hi/T2lo T2 weighted, high/low resolution.
  • SWI Susceptibility weighted.
  • DSC Dynamic susceptibility-weighted perfusion (HOF_Id=DSC)
  • DWI Diffusion tensor imaging (HOF_Id=DWI)

NOTES:
1. Multiple DSC/DWI scans are not supported. Only the first of each type will be registered.
2. Perfusion and diffusion derived scans (e.g. FA), whether local or imported, will be coregistered to the first DTI/DSC scan encountered by the pipeline.

Outputs

The following will result from a successful completion of the pipeline

  • The user who launched the pipeline will receive an email from the XNAT Team with the wording:
    "HOF processing completed for..."
  • Files will be created in the Resources directory for the scan under the folders HOF_QC and HOF_reg
  • HOF_QC contains QC pdf, snapshots and pipeline configuration files, as well as 4dfp images in atlas space
  • HOF_reg contains 4dfp images in original space and coregistered to the target T1w image.

    Screenshot of directories
    HOFfiles.jpg

Troubleshooting

The email message informing of pipeline failure will have *.log and *.err files attached. Scroll to the very end of the email to see the error message marked by "ERROR" text).

Common issues
Problem: "REGISTRATION FAILED" message appears for scan X in the QC pdf.
Possible cause 1: structural image header information was missing/not extracted correctly.
Solution: Check the .ifh header file for that image converted to 4dfp.
Possible cause 2: this is not the first diffusion/perfusion scan. Multiple physiological sequences of the same type are not supported within one HOF session.
Solution: Choose only one perfusion and/or diffusion scans per HOF session.

Problem: Perfusion or diffusion processing fails.
Possible cause 1: T2w scan is not coregistered well.
Solution: Try another T2w scan with targ_t2, or, if not available, set targ_t2 to a high resolution T1w scan.
Possible cause 2: DTI directions could not be inferred from the raw data and/or could not find specified direction file.
Solution: Try custom_dti=1 option, or choose manual direction file using dti_prm.
Possible cause 3: DTI encoded as [Siemens] mosaic was not decoded correctly due to wrong/missing slice dimensions in DICOM header.
Solution: manually set the size of DTI slice (dwi_nx, dwi_ny)

Problem: Registration of target image to atlas (always first in the QC PDF) is unsatistactory.
Possible cause 1: partial brain field of view (FOV) coverage, or high gain field inhomogeneity in registration target.
Solution: choose a specific T1w target (set targ_t1) with better FOV/uniformity.
Possible cause 2: poor anatomic match between atlas and target T1w image
Solution: set targ_t1 to a different high resolution T1 weighted image (T1hi or MPRAGE) or choose specific target atlas (atlas variable).

Problem: Registration of perfusion or diffusion sequence to T1w target is unsatisfactory.
Possible cause 1: partial FOV coverage of DTI/DSC scan; high EPI distortion of DTI/DSC.
Solution: if possible, choose another DTI/DSC acquisition.
Possible cause 2: automatic detection of DTI/DSC encoding of z direction failed (registrations look z flipped).
Solution: set dwi_zflip/perf_zflip to 1.

References

Milchenko M, Snyder AZ, LaMontagne P, Shimony JS, Benzinger TL, Fouke SJ, Marcus DS. Heterogeneous Optimization Framework: Reproducible Preprocessing of Multi-Spectral Clinical MRI for Neuro-Oncology Imaging Research. Neuroinformatics. 2016. doi: 10.1007/s12021-016-9296-7. PubMed PMID: 26910516.