Hi Jason,

  I developed the DICOM file reading in ChimeraX and Elaine shared with me the 1000 dicom images you sent her.  ChimeraX does not understand that this is 25 3D stacks each 40 planes.  Probably that data is in the DICOM files and we are just not handling the case of a 3D time series provided all as 2D slices -- maybe we only handle when each time point 3D stack is in a single DICOM file.  It opens it as a single 3D image of all 1000 planes.  It would take more study to figure out what ChimeraX accepts.  The DICOM standard is many thousands of pages so we only handle a very limited subset so far.  But we are trying to expand what it can handle.

  I made a movie of your data, but as Elaine suggested it was tedious to setup.  First I split the 1000 files into 25 directories, 40 2-D images per directory, then I opened them with ChimeraX File / Open DICOM Folder.  Then I could play through them with the ChimeraX mseries command.  I could show full 3D volume rendering or planes or a thin slab.  The thin slab seemed the best view.  ChimeraX somehow extract a weird rotation matrix from the files that broke the orthoplanes display so I set the 25 3D images to have no rotation.  I saved all the images in one Chimera map file (suffix .cmap) -- this reads in a lot faster.  Below is an image of what it looked like.  I recorded a movie but I'll send that separately to you in case you don't want others to see it.

  ChimeraX is not ready to easily handle this DICOM time series data.  But we will look into improving it using your example file as a (private) test if that is ok.  We'd love to have a collaborator who can give us directions on what you want to see and give us example data.  We are specifically interested in virtual reality visualization of DICOM data in collaboration with NIAID.

  I wrote a Python script to split your 1000 dicom files into 25 directories and another to zero all the rotation matrices, and those scripts can be found on a ChimeraX enhancement request I made

https://www.rbvi.ucsf.edu/trac/ChimeraX/ticket/6255

  Tom



On Feb 25, 2022, at 2:35 PM, Elaine Meng via ChimeraX-users <chimerax-users@cgl.ucsf.edote:

Hi Jason,
Well, I didn't get too far.  Gory details:

- simply renaming the .dcm files to an image filename suffix like .tif did not work, as the files are still somehow marked as DICOM.  They are binary (not human-readable in text editor) so beyond my skill set to understand their contents.

- our DICOM reader requires specifying the whole folder, and does not allow only reading a subset of files in the folder.  So I tried putting images 1-40 in one folder and a 41-80 in another folder, as a small test.  I could open each folder in turn which gives two volumes open in ChimeraX, corresponding to the first 2 time steps. Each looks correct when shown as "plane" or "box" in Volume Viewer, although the orthoplanes outlines still look skewed and somehow wrong.

- unfortunately volume models can't be grouped into a series after the fact.  Instead one has to specify multiple volume data files as a series when they are opened.  So I can imagine a very tedious workflow in which each timestep is a separate DICOM folder containing 40 planes. You would open each folder, then save a volume data file in some other format.  Eventually you'd have 25 volume data files which you could then open en masse with "open" and the "vseries true" option.

Now I'm not suggesting any of the above is actually feasible for any normal human being to do... it is just the only workflow I can think of currently that might possibly work given current limitations of ChimeraX.

Another limitation may be the information in the data itself.  We do have at least a couple of DICOM examples containing time series that do open correctly as time series.  It may be that your files do not have the metadata needed to specify that they comprise a time series, or if they do, it is in a different way than ChimeraX expects.

The working examples we have of DICOM time series are:

(1) Mouse-Astrocytoma collection, patient ID TVD_GBM_IC1_070610_16_100110, study date 2010-10-01
<https://wiki.cancerimagingarchive.net/display/Public/Mouse-Astrocytoma">

(2) time series example on 3D slicer wiki
<https://wiki.slicer.org/slicerWiki/images/c/c2/DCE_series.zip>

I'm not sure whether there is enough info in your files that we could make ChimeraX recognize them as 25 stacks of 40 planes each rather than one big stack.  However, it is very useful that you reported these issues so that we can try to make the DICOM reader better able to handle more situations, if possible.  If we make progress on that, will let you know.

In the meanwhile, apologies for the limitations!!
Best,
Elaine
-----
Elaine C. Meng, Ph.D.                       
UCSF Chimera(X) team
Department of Pharmaceutical Chemistry
University of California, San Francisco




On Feb 25, 2022, at 1:18 PM, Rubenstein, Jason via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:

Thanks for looking into it.  I did export it as anonymous so hopefully no PHI in the dicom headers.


Jason Rubenstein, M.D., F.A.C.C., F.H.R.S.
Vice Chief, Electrophysiology
Codirector, Cardiac MRI
Associate Professor, Medical College of Wisconsin
Office: 414-955-6777

From: Elaine Meng <meng@cgl.ucsf.edu>
Sent: Friday, February 25, 2022 3:07 PM
To: Rubenstein, Jason <jrubenstein@mcw.edu>
Cc: chimerax-users@cgl.ucsf.edu <chimerax-users@cgl.ucsf.edu>
Subject: Re: [chimerax-users] Stack thickness question and cine playing?

ATTENTION: This email originated from a sender outside of MCW. Use caution when clicking on links or opening attachments.
________________________________

Hi Jason,
Ah, I hadn't realized from the initial question that this was DICOM data, although of course it makes sense given your area of research.  There are likely some ChimeraX bugs and misinterpretations reading in the data, as the DICOM functionality has not been as thoroughly exercised as features for working with atomic density maps.

So I see 1000 planes, and I think they are just all getting put into one big stack instead of 25 separate stacks (time points). If so, it might be fixable by changing all the filenames so that they reflect time point, and end in image file suffix (e.g. .tif if that's what the DICOMs contain)... but I would have to investigate further and get back to you. Maybe there is metadata in each of the .dcm files that marks the different time points and slice numbers, and if so, it would be nice to improve our DICOM reader to understand this so that renaming would not be required.

I still don't know why the orthoplanes axes don't seem aligned, however.  ChimeraX may just be confused about this data.

I will share internally with a few experts on our team with the understanding that this data should remain private, and we will get back to you with any solutions we find.

Sorry about the difficulties -
Elaine

On Feb 25, 2022, at 10:50 AM, Elaine Meng via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:

Hi Jason!

With (1), I'm not sure what is happening... from the orthoplanes image you sent, one guess is that the 3 axes of the data are not along X,Y,Z and so a larger XYZ box is shown.  Or maybe the slices are placed too far apart, which may be defined (or misinterpreted from) the input files.  Or if you simply meant that the scale of the data is wrong, that can be reset with the "volume" command "voxelSize" option.

I understand that this may be private data, but if there is example data you can share, it would help in defining and possibly solving this problem.  You could send it to just me rather than chimerax-users if that helps.

(2) If the data are recognized as a volume series, then the series can be played with a slider or the "vseries" command.  What kinds of data can be recognized and how to specify that are described here:
<https://urldefense.com/v3/__https://rbvi.ucsf.edu/chimerax/docs/user/trajectories.html__;!!H8mHWRdzp34!s0-lHURJ9t1XwqkJL13HezYzErSctPMhHjQSds6R4BaEHrGoEjJ5LLeFag45_rn7hA$ >

I hope this helps,
Elaine
-----
Elaine C. Meng, Ph.D.
UCSF Chimera(X) team
Department of Pharmaceutical Chemistry
University of California, San Francisco

On Feb 25, 2022, at 9:16 AM, Rubenstein, Jason via ChimeraX-users <chimerax-users@cgl.ucsf.edu> wrote:


Hi!  Two questions:

    • This 3D dataset is only about 30 mm thick in the Z direction.  When ChimeraX is displaying it, it shows the thickness about 50x too thick.  Any way to adjust this?
<image.png>

    • this is a "4D" dataset, with 20 phases of 3D stacks.  Any way to "play" the cine?
Thanks!

Jason Rubenstein, M.D.
Codirector, Cardiac MRI
Associate Professor, Medical College of Wisconsin


_______________________________________________
ChimeraX-users mailing list
ChimeraX-users@cgl.ucsf.edu
Manage subscription:
https://www.rbvi.ucsf.edu/mailman/listinfo/chimerax-users


_______________________________________________
ChimeraX-users mailing list
ChimeraX-users@cgl.ucsf.edu
Manage subscription:
https://www.rbvi.ucsf.edu/mailman/listinfo/chimerax-users