Hi,
I am trying to run the example "Plot the results of a segmentation with SPM-based pipeline T1xT2", but there is an error in the line "data_path = load_test_data("data_test_macaque_processed")". It says : AssertionError: Error, data_zip = C:\Users\dell\data_macapype\data_test_macaque_processed.zip not found .
I checked the C:\Users\dell\data_macapype folder and fount it empty. I also try to run other examples and met the similar error when trying to load the example data.

I would hope to ask where to find or to download the example data. Thank you very much!

As we are now extending the package to further NHP species (including so macaque, marmoset, baboon and more recently chimpanzee, the number of corresponding params file are increasing; in fact , in extra to template choise, only some nodes are critical for each species (in particular, atlas_brex parameters in ANTS pipelines); most of the other are similar.

The idea would be to keep the -params but makes it optionnal, and add a new -species, with predefined keyword (i.e."macaque", "marmoset", "baboon", "chimpanzee") that would take a default params file, already created (most of them already exists).

Re: our Jan 21 meeting here are two examples of the atlas brex documentation. I think the piece that is missing is a listing of the actual params, and roughly what they do, and a default setting. Everything I added for this I grabbed from either the AtlasBrex class or the bash script itself.

Current:

inputnode:
restore_T1: preprocessed (debiased/denoised) T1 file name
restore_T1: preprocessed (debiased/denoised)T2 file name

arguments:
params_template: dictionary of info about template
params: dictionary of node sub-parameters (from a json file)
name: pipeline name (default = “extract_pipe”)

Outputs:
smooth_mask.out_file: Computed mask (after some smoothing)

Proposed:

inputnode:

• restore_T1: preprocessed (debiased/denoised) T1 file name
• restore_T1: preprocessed (debiased/denoised)T2 file name

arguments:

• params_template: dictionary of info about template
  • template_head: Atlas with skull default: NMT
  • template_brain: Atlas with skull removed, default NMT_SS
• params: dictionary of node sub-parameters (from a json file)
  • f: initial BET fraction, default 0.5
  • reg: method of template registration, default 1
    0 = FNIRT w/ bending
    1 (default) = FNIRT membrane-energy regularization
    2 = ANTs/SyN w/ N4Bias
    3 = w/o N4Bias
  • w: warp resolution, default 10,10,10
  • msk: mask binarization threshold (in %) for fslmaths default a,0,0
• name: pipeline name (default = “extract_pipe”)

Outputs:

• smooth_mask.out_file: Computed mask (after some smoothing)

How to add a new parameter?

How to create your own pipeline?

Hi!
You should change PNH to NHP here: https://macatools.github.io/macapype/
;)

For visualization with mrtrix, zeros are not understood as background so it is not possible to use the 3D view anymore - everything is inside a gray box-. If replaced by NaNs, the different tissues could be visualized in 3D using mrtrix.

Rather than implementing 3drefit -deoblique on the native image, to implement 3dWarp -card2oblique to convert the transformed template masks (non-oblique) with the same oblique form as the native image.

DEBUG avec remi:

• probleme avec le git clone de nifti reg ?
• replace ?/nodes/../bash par ?/bash directement dans le node AtlasBREX

segment_pnh -data inputs/BIDS_raw -out outputs -sub 001 -ses 001 -soft ANTS -params code/params_segment_macaque_ants.json -deriv -pad -nprocs 2

produces the following:

outputs/derivatives/macapype_orig_ants
└── [   3]  sub-001
        └── [   3]  ses-001
            └── [   3]  anat
                └── [   3]  datasink
                    └── [   3]  outputs
                        └── [   3]  derivatives
                            └── [   3]  macapype_orig_ants
                                └── [   3]  sub-001
                                    └── [   3]  ses-001
                                        └── [  10]  anat

full paths get it right:

segment_pnh -data $(pwd)/inputs/BIDS_raw -out $(pwd)/outputs -sub 001 -ses 001 -soft ANTS -params code/params_segment_macaque_ants.json -deriv -pad -nprocs 2

outputs/derivatives/macapype_orig_ants
└── [   3]  sub-001
        └── [   3]  ses-001
              └── [  10]  anat

For portable setup of the computational environment, relative paths are essential.
Great tool, keep up the great work!

There are new versions of the NMT template and scripts (e.g. NMT subject align). To incorporate into Macapype.

https://github.com/jms290/NMT

Re: our meeting I am putting this here so it is in a central place. There are two reasons that freesurfer and macapype may not play nice together.

There are two freesurfer issues where it changes data in disadvantageous ways:

1. run with the tag -hires or it will always resample to 1m iso. That is a recon-all tag so it is not in all binaries but most.
2. freesurfer puts things in its freesurfer space (which leaves them in RAS in other viewers like itksnap).
   For surfaces the command mris_convert with the --to-scanner flag will apply the correct transform to fix this.
   For volumes it is the transform given by mri_info --vox2ras-tkr orig.mgz I think but I am less sure.

I wonder if we should not create and output folder where all the useful images will be: brain mask, segmentation tissue probability images, surfaces ... It is cumbersome to navigate through all the directories created by nipype. It is ok for debugging, not for production. If I take the example of fmriprep, which paves the way, the nipype outputs go into a temporary folder and the usable output is saved in the derivatives/fmriprep folder of the project.

Hi, thanks for making this package!

I successfully installed macapype via Docker and tried to run it using the instructions provided here.

Unfortunately, /opt/packages/macapype/workflows/ used in the commands does not exist in the container. I am aware that the segment_pnh command can be used instead (see n.b. here), however, params_segment_pnh_ants_based.json cannot be found. I don't know if I'm missing something or if there is an error in the instructions. Any guidance here would be helpful! Thanks

As suggested by @FeHoff

Hi, browsing the repo, I see that there's no software license defined for macapype... I think it'd be good to define one ;)

BIDS standard does not require a session for single session data. The code create_datasource currently assumes an iterable across sessions and this needs to be made optional to be the same as bids standard.

Reorient pipeline sends the image on at the swap_dim node not the reorient node. This iss fixed in ostanley/macapype:7T and I will PR it in soon.

Keep the reg aladin + NMT_subject_align / animal_warp linear and non linear transfo, as well the inverse trans, in the derivatives + renamed properly

pipeline with T1 only file. Modify the correct bias by debias N4 → leads to new functions with _T1 in the name ?

So far, even if macapype exists on pypi, there is no real way to launch macapype except from cloning the github repo and
$ python workflows/segment_pnh.py ...
(the workflows directory in not packaged yet, neither is the segment_pnh.py)
There should be solutions with the equivalent of an executable whose location could be added to the PATH once installed; could be related to issue #129, the params.json files are not involved in the package either

There is an indentation error/minor typo in _create_prep_pipeline which means you cannot add reorient to it. It has been updated in my fork and I will PR when I confirm it works.

As we are now extending the package to further NHP species
(including so macaque, marmoset, baboon and more recently chimpanzee,
the number of corresponding params file are increasing; in fact , in
extra to template choise, only some nodes are critical for each species
(in particular, atlas_brex parameters in ANTS pipelines); most of the
other are similar.
The idea would be to keep the -params but makes it optionnal, and add
a new -species, with predefined keyword (i.e."macaque", "marmoset",
"baboon", "chimpanzee") that would take a default params file, already
created (most of them already exists).

Inspired from subject sub-032159 in site-mountsinai-S from PRIME-DE, which fail miserably since the first step of macapype.

Here is for instance the image after the first stage:

• option 1
  -- Prepare the data with FSL-FAST:
  fast -l 8 -B sub-032159_ses-001_run-1_T1w.nii.gz
  -- Apply the reg_aladin command used in macapype to the bias corrected image:

reg_aladin -aff sub-032159_ses-001_run-1_T1w_aff.txt -flo sub-032159_ses-001_run-1_T1w_restore.nii.gz -ln 12 -lp 10 -nac -noSym -omp 1  -ref /Users/jsein/Documents/Centre_IRMf/DATA/BIDS/Covoprim/inia19-t1.nii -res sub-032159_ses-001_run-1_T1w_restore_res.nii.gz -rigOnly -smooR 1.000000

-- Apply the calculated transform to the original image

reg_resample -flo sub-032159_ses-001_run-1_T1w.nii.gz -ref /Users/jsein/Documents/Centre_IRMf/DATA/BIDS/Covoprim/inia19-t1.nii  -res test.nii.gz -trans sub-032159_ses-001_run-1_T1w_aff.tx

This is the result:

• option 2 (much faster)
  -- change the parameters of reg_aladin on the original image:

reg_aladin -aff sub-032159_ses-001_run-1_T1w_aff.txt -flo sub-032159_ses-001_run-1_T1w.nii.gz -ln 8 -lp 10 -nac -noSym -omp 1  -ref /Users/jsein/Documents/Centre_IRMf/DATA/BIDS/Covoprim/inia19-t1.nii -res sub-032159_ses-001_run-1_T1w_res.nii.gz -rigOnly -smooR 1.000000

=> same result

@kepkeeloh suggests to check how the T2w after debiased works instead of the T1w

Even better = all nodes can be params/indiv_params (requires a full wrap) and provide the json all available options

Several users, in particular using pediatric data, requested to provide local template; it should be taken into account that their is now two types of tissues priors provided, one with 3 different files for csf/gm/wm (proba_priors), and one with index based tissues in one file

-IsoSurface (AFNI) is bugging as libXp6.so is missing, fixed in skullTo3d but would make more sense to add it in the macapype_env directly

• the clean up is not properly set (not a big problem)

Cannot access to the version number from the python console.

Integrate the JunaChimp Template and DaVi parcellation for chimpanzee mapping and cross species comparison into mamapype

As Reguired by @julfou81

multi_params should not break with assert if a node params is not in the multi_params
but return empty dict and keep with the default from params

https://afni.nimh.nih.gov/pub/dist/doc/htmldoc/nonhuman/macaque_tempatl/template_nmtv2.html#nh-macaque-template-nmtv2

Download:
wget https://afni.nimh.nih.gov/pub/dist/atlases/macaque/nmt/NMT_v2.0_sym.tgz
tar -xvf NMT_v2.0_sym.tgz

In extra to the minimalist pipeline nii_to_mesh provided so far, some users would require a pipeline inspired by what is done in NHP-freesurfer:
https://github.com/VisionandCognition/NHP-Freesurfer/blob/public/Notebooks/NHP_Surfaces_and%20Flatmaps.ipynb

Sometimes, the T2-weighted images might give better contrast for the WM/GM segmentation, i.e. for very young and poorly myelinated brains.

So it would be interesting to adapt macapype to have an option to perform segmentation using T2-weighted images.

• make sure the docker are working properly (both Dockerfile and Docker pull version)
• use the docker to create proper unit-test