Feature request: Support for ASL

[ChristianNSchmitz]

Dear nipy-team,

We would like to ask you if you are planning support of heudiconv for ASL (newly published on BIDS v1.5.0 on 2021-02-23)?

Best wishes and thanks! Christian

[yarikoptic]

yeah, would be nice... never dealt with that beast yet so have no clue on its "specifics" -- any sample dicoms available anywhere. I wonder what special treatment would it need if any? i.e. I see that some specific heuristics (like reproin) might need to be adjusted but heudiconv code itself might not need?

[ChristianNSchmitz]

It might make sense to ask the authors of the bids-standard for dicoms. They have example data for the bids dataset (https://drive.google.com/drive/folders/13mqn_XZC5rfICJwXT0QSuVoR-wWh3k--). Unfortunately, we use a custom sequence that doesn't include most of the necessary information in the dicom-header..

[neurolabusc]

My response is similar to issue 514, except BEP005 instead of BEP009. The files you link to only show the BIDS/NIfTI solution, without the DICOM images that they should be derived from. Without exemplars of input and output, it is hard to support this.

I did generate my own ASL data and attempt to comprehensively extract all the ASL information I can from these datasets. However, I only have Siemens equipment, and the data stored by Siemens is very different from what BEP005 expects. I think the Philips and GE ASL-specific DICOM tags are even more limited. It seems that BEP005 was developed based on what information is required to process this modality, not the information provided in the DICOMs for this modality. Therefore, users will need to complete the JSON files themselves.

[neurolabusc]

The latest development release of dcm2niix has improved support for ASL data for GE, Philips, and Siemens. All relevant data from the DICOM header is now stored in the BIDS format JSON sidecar. Unfortunately, for GE and Philips in particular there is very little data stored in the DICOM images, so a user must supply details from other sources. I would encourage users with equipment from GE and Philips to lobby their manufacturer to provide more details in their DICOM data.

Here are some validation datasets:

-Philips classic DICOM dcm_qa_philips_asl -Philips enhanced DICOM dcm_qa_philips_asl_enh

• Siemens dcm_qa_asl

I suggest we close this issue, this is as far as we can go with the data provided.

[yarikoptic]

@HenkMutsaerts mentioned that some metadata might be available in Phoenix reports for Siemens (IIRC support was added in #489 and may be elsewhere), and in XX files for Philips. They have developed a library which already does that "post-treatment" to dcm2niix converted files https://github.com/ExploreASL/ExploreASL/blob/main/Modules/SubModule_Import/xASL_wrp_NII2BIDS.m . ATM can't be reused as is due to license constraints but things might change.