dcm2niix with resting-state multiband data

[mm4355]

I want to convert dicom MUX 6 to .nii.gz

I attached a picture on the problem when using dcm2niix, It seams dcm2niix needs a guide on how to read the headers... not sure if I am doing it correctly. Please advice on how to use dcm2niix with multiband resting data.

Thanks

[neurolabusc]

This appears to be a replication of issue 220. I assume these are DICOM images from a GE scanner running the older WIP rather than the recent GE product hyperband sequence. With the WIP package the recon is not computed on-line.

We do not have GE hardware at my center, but Gareth Barker at King's College London described this to me as:

The WIP saves the raw (k-space) data (as a GE proprietary "Pfile") for later offline reconstruction, but also does a (non-multiband aware) online recon which will created the aliased images you're seeing. The online recon images are really only intended to a) allow the user to check something is actually running and b) provide a full, GE standard, DICOM header that can then be used during the offline recon. (The offline recon (at least in the form I've seen it) takes a single DIOCM image from the online recon, plus the raw data, as inputs). This means that: a) I don't think there is any way to recover the full dataset from these aliased images alone, unfortunately, on the console or otherwise and b) even if the user can find and reconstruct the original raw data, the DICOM headers should be treated with a degree of caution. (The offline recon correctly updates things to ensure unique UIDs, etc, but I'm not sure every tag necessarily gets correctly updated to allow for the 3-fold increase in slice numbers.

If I am correct, you will need to reconstruct from your P files.

[mm4355]

Thanks for your response. Yes, these are DICOM images from a GE scanner but we are running the recent GE product hyperband sequence. Please advice on how or what to do.

Sent from my iPhone

On Sep 26, 2018, at 11:58 AM, Chris Rorden notifications@github.com wrote:

This appears to be a replication of issue 220. I assume these are DICOM images from a GE scanner running the older WIP rather than the recent GE product hyperband sequence. With the WIP package the recon is not computed on-line.

We do not have GE hardware at my center, but Gareth Barker at King's College London described this to me as:

The WIP saves the raw (k-space) data (as a GE proprietary "Pfile") for later offline reconstruction, but also does a (non-multiband aware) online recon which will created the aliased images you're seeing. The online recon images are really only intended to a) allow the user to check something is actually running and b) provide a full, GE standard, DICOM header that can then be used during the offline recon. (The offline recon (at least in the form I've seen it) takes a single DIOCM image from the online recon, plus the raw data, as inputs). This means that: a) I don't think there is any way to recover the full dataset from these aliased images alone, unfortunately, on the console or otherwise and b) even if the user can find and reconstruct the original raw data, the DICOM headers should be treated with a degree of caution. (The offline recon correctly updates things to ensure unique UIDs, etc, but I'm not sure every tag necessarily gets correctly updated to allow for the 3-fold increase in slice numbers.

If I am correct, you will need to reconstruct from your P files.

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[neurolabusc]

Are the images created by dcm2niix identical to what you see when you look at the images with a DICOM viewer (e.g. Horos). If so, the issue is as I described previously. If they are different, can you share a dataset with me?

[mm4355]

Attachment available until Oct 27, 2018 I want to share this dataset with you, please find attached 1 resting-state fMRI run from a test subject.

Thanks for your help.

Click to Download https://www.icloud.com/attachment/?u=https%3A%2F%2Fcvws.icloud-content.com%2FB%2FAS71ZY-bGujz0NBkMK-2JlMoa-Q_AWdrJVKJ1mwcvXfvMYESQjRALK0z%2F%24%7Bf%7D%3Fo%3DAn5LTgNz4C17OXCWV3u6qvR3j6-Fl6YRPVviFqlZzURn%26v%3D1%26x%3D3%26a%3DB8aPFhW6utfRaGAsM56HIl6-_XFKA2nJRgEACAHIAP84Z-MsA6kgRg%26e%3D1540648814%26k%3D%24%7Buk%7D%26fl%3D%26r%3DEDDAF24E-1423-402A-909A-5A81529DFCB8-1%26ckc%3Dcom.apple.largeattachment%26ckz%3D92459E3B-D997-4FEF-B828-EAE0127F506F%26p%3D16%26s%3DDOBykxVV5RL4v5am_O-e-1L8cSo&uk=BPgYhx2E8pHR7losgNXivA&f=s000009.zip&sz=336614837s000009.zip 336.6 MB

---

Leah Moreno, PhD Research Scientist Division of Experimental Therapeutics Department of Psychiatry Columbia University Medical Center 1051 Riverside Drive, Unit 21 New York, NY 10032 phone: (914) 218-7311 email: mm4355@cumc.columbia.edu mailto:mm4355@cumc.columbia.edu

On Sep 26, 2018, at 5:22 PM, Chris Rorden <notifications@github.com mailto:notifications@github.com> wrote:

Are the images created by dcm2niix identical to what you see when you look at the images with a DICOM viewer (e.g. Horos https://horosproject.org/). If so, the issue is as I described previously. If they are different, can you share a dataset with me?

— You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/rordenlab/dcm2niix/issues/233#issuecomment-424874213, or mute the thread https://github.com/notifications/unsubscribe-auth/Apl9oGk3nCrdEYcnLYQmh5LaAdAPjctBks5ue_ABgaJpZM4W66lC.

https://assets-cdn.github.com/images/email/message_cards/header.png https://assets-cdn.github.com/images/email/message_cards/avatar.png https://github.com/rordenlab/dcm2niix https://horosproject.org/ https://github.com/rordenlab/dcm2niix/issues/233#issuecomment-424874213 http://schema.org/ https://github.com/rordenlab/dcm2niix/issues/233#issuecomment-424874213 https://github.com/rordenlab/dcm2niix/issues/233#issuecomment-424874213 https://github.com/ http://schema.org/extensions https://assets-cdn.github.com/images/email/message_cards/avatar.png https://api.github.com/ https://api.github.com/ https://github.com/rordenlab/dcm2niix/issues/233#issuecomment-424874213 https://api.github.com/

[neurolabusc]

I can not replicate your issue. These files seem to convert well with dcm2niix v1.0.20180918 and displayed using MRIcroGL, FSLeyes, FSLview, etc. Can you check the version of dcm2niix you are using and also using one of the display tools I used. For Linux and Mac, you can build the latest code from the script below. For Windows computers you can click on the green "AppVeyor" icon on the dcm2niix Github home page, then click the "artifacts" button to get a compiled copy of the most recent commit.

$ git clone https://github.com/rordenlab/dcm2niix.git
$ cd dcm2niix/console
$ make
$ ./dcm2niix
Chris Rorden's dcm2niiX version v1.0.20180918  GCC6.1.0 (64-bit MacOS)
...

[mm4355]

Attachment available until Oct 27, 2018 I was expecting an image like the attached, I can not see full brain in the parasagittal or coronal view, I am now using the script you sent for Mac.

Here I have attached a .nii file converted from p-files using the old multi band from GE. So was expecting the same output when converting DICOMS to nii using the hyperband sequence. Am I wrong?

Thanks a lot.

Click to Download https://www.icloud.com/attachment/?u=https%3A%2F%2Fcvws.icloud-content.com%2FB%2FASR-DkoQFs3AVh5stDmG_8VSM4vAASTj8On_lCorxSOUMPxH6-EyYPYK%2F%24%7Bf%7D%3Fo%3DAvqHmNvgq8FJ_r42hfF3Zl0rSwdAxzutMS_0ibcqnKSu%26v%3D1%26x%3D3%26a%3DB3X4HMjLY2j76X5a6eB0LZJE3zxxA83JRgEACAHIAP9uuPK6Aw0hRg%26e%3D1540654816%26k%3D%24%7Buk%7D%26fl%3D%26r%3D5EE0BC18-D62C-4B72-A4C3-16C1F4F30510-1%26ckc%3Dcom.apple.largeattachment%26ckz%3D92459E3B-D997-4FEF-B828-EAE0127F506F%26p%3D16%26s%3DHsPNm-8B5GoEjyqRuyEMW2HdyNk&uk=h_ypPmetXPH3r9Z_L5fnNw&f=E15386_S11_mux6_arc1.nii.gz&sz=1519851386E15386_S11_mux6_arc1.nii.gz 1.52 GB

---

Leah Moreno, PhD Research Scientist Division of Experimental Therapeutics Department of Psychiatry Columbia University Medical Center 1051 Riverside Drive, Unit 21 New York, NY 10032 phone: (914) 218-7311 email: mm4355@cumc.columbia.edu mailto:mm4355@cumc.columbia.edu

On Sep 27, 2018, at 10:45 AM, Chris Rorden <notifications@github.com mailto:notifications@github.com> wrote:

./dcm2niix

[mm4355]

using wb_view compare both images, the full brain one is converted from p-files using the old GE multiband sequence, the other one is converted from DICOM using the hyperband GE sequence with dcm2niix.

---

Leah Moreno, PhD Research Scientist Division of Experimental Therapeutics Department of Psychiatry Columbia University Medical Center 1051 Riverside Drive, Unit 21 New York, NY 10032 phone: (914) 218-7311 email: mm4355@cumc.columbia.edu

On Sep 27, 2018, at 10:45 AM, Chris Rorden notifications@github.com wrote:

I can not replicate your issue. These files seem to convert well with dcm2niix v1.0.20180918 and displayed using MRIcroGL, FSLeyes, FSLview, etc. Can you check the version of dcm2niix you are using and also using one of the display tools I used. For Linux and Mac, you can build the latest code from the script below. For Windows computers you can click on the green "AppVeyor" icon on the dcm2niix Github home page, then click the "artifacts" button to get a compiled copy of the most recent commit.

$ git clone https://github.com/rordenlab/dcm2niix.git $ cd dcm2niix/console $ make $ ./dcm2niix Chris Rorden's dcm2niiX version v1.0.20180918 GCC6.1.0 (64-bit MacOS) ...

https://user-images.githubusercontent.com/8930807/46153757-c5ae5700-c241-11e8-93f7-3e021e1a14cb.jpg — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub https://github.com/rordenlab/dcm2niix/issues/233#issuecomment-425120406, or mute the thread https://github.com/notifications/unsubscribe-auth/Apl9oKQDaVFpsxsO7E_pe58Kuxc5_bNCks5ufOSDgaJpZM4W66lC.

[neurolabusc]

dcm2niix faithfully converted the DICOM images it was provided.