Loading of GE/Kretz ultrasound volumes (.vol file)

Thank you fro the sample files. I was able to load the files manually but still need to figure out how to get some metadata from the files automatically.

Getting some more files, some of them showing objects with known size would help.

I can do… We will do a a study and supply measurements on them… But how did you get them to load manually?

I’ve compared the content of each file and from the differences determined where voxels and image size information is stored. Now the challenge is that voxels are stored in a rectangular grid but the image is acquired with a convex probe, so I need to figure out where the spherical projection parameters are stored. For this, it would help if you could also acquire a scan of a flat surface, such as a bottom of a water-filled plastic box, at a few different depths.

At what levels do you need the measurements?..

Basically the tech wants to know what you need and she will scan it for
you…

Thanks

Something like these:

  • probe imaging depth set to 5cm, have the water tank bottom at 3cm
  • imaging depth set to 10cm, have the water tank bottom at 5cm
  • imaging depth set to 10.5cm, have the water tank bottom at 5cm
  • imaging depth set to 11cm, have the water tank bottom at 5cm
  • imaging depth set to 15cm, have the water tank bottom at 10cm

Always have an object of known size in the image (such as a dice or coin).

In the meantime, I’ve found out how data fields are stored in the image header, I can retrieve angular resolution, image size, field of view, etc. The only information that I could not decode yet is the imaging depth start/end. That’s why the above described set of images with different imaging depth would help.

Importing is now available in latest Slicer nightly version, in SlicerHeart extension. See a demo here:

Conversion of spherical coordinates to Cartesian is not perfect yet, there are distortions in the images. If you send me those additional images that I’ve asked then there is a good chance I can fix all remaining issues.

Andras, did you already get the data you need?

Amir

I would still need some more data to finalize scan conversion (from spherical to Cartesian). Put a small object of known dimensions (such as a dice or thick coin) into water container with flat bottom and acquire images like these:

  • start imaging depth set to 0cm, imaging depth set to about 10cm. This setting will serve as baseline, keep all parameters that are not mentioned unchanged compared to these settings.
  • imaging depth set very slightly (by 5-10mm) less
  • imaging depth set very slightly (by 5-10mm) more
  • imaging depth set to much more (by 3-5cm)
  • imaging depth set to much less (by 3-5cm)
  • field of view decreased a bit (by a few degrees)
  • field of view decreased significantly (by at least 15-20 degrees)
  • start imaging depth set to a small value (1-2cm?)
  • start imaging depth set to a larger value (3-5cm?)

I will try to obtain the scans till the end of the week. If it helps I will also try take the object to the angio suite or CT so that you have an exact 3D representation for comparison.

Moreover, there was a plugin for Amira written at our institution years ago to import the kretz files. Unfortunately the developer left and I cannot find the source code. All I have now is the dll, that you could disassemble to get to the algorithm. But I don’t think thats helpful. I will try to search for the source code though.

Amir

Thanks in advance.

If the Amira reader implemented scan conversion then it’s source code would be very useful.

This is a “brainlab disposable reflective marker sphere” in a cylinder container with water. The bottom of the container is not perfectly flat. I didnt find any better Phantom. There Voluson E8 exports also DICOM, DICOM with DICOMDIR, 4dv (i took uncompressed, proprietary format again I think).

For the vol files, it exports vol (nicht cartesian), vol with wavelet (compression I guess, can be lossy or not lossy, so I took the not lossy) and most imortantly, VOL Karthesisch=cartesian.

So I think maybe modifying your algorithm so that the volumes of carthesian and not carthesian match could help.

I have given up the search for the source code of our amira plugin, because its long lost and as far as I know, it only imports the “cartesian” volumes.

What I want to say is that there are also cartesian volumes around, that could be imported with more ease. However, I don’t know since what version the export of cartesian is supported on the machines.

Let me know if this helps or is completely useless and I should search for another phantom.

Amir

1 Like

Thank you these images are very useful. I’ll try to update the code to handle Cartesian volumes and improve scan conversion of spherical volumes.

hi guys, any progress?

hey @lassoan , i am a student (Computer Science) from an university from Chile (Universidad Austral de Chile , UACH), and i am part of a group that is trying to read Kertz ultrasound to make a 3D impression.
We are thinking if you need help with more data sets or even development, to improve the scan conversion, because we might be able to help you if you want, we have a Voluson i if you are interested.

2 Likes

It would be great if you could help with improving the importer. There are two main limitations:

  • import from simple rectilinear volumes is not implemented yet (we would just need to figure out which fields store image spacing and use that and disable scan conversion)
  • it is not clear how to determine start/end radius for spherical acquisitions (I’ve got some more data that should help in finding out which fields to use and how, but I did not have time to analyze it yet)
1 Like

I followed the instruction that you suggest but have no success to load .vol imagens in 3D Slicer. This .vol imagens opens in TomoVision but not in 3D Slicer. Any advice?

1 Like

I have only implemented importing of images acquired with a spherical probe, your might have been acquired with a cylindrical, or there could have been other differences. I’ll improve the importer in the following weeks and post an update here.

1 Like

I am having the same problem as Charlie and the probe for the GE machine is a rab6-4d convex probe. I’m pretty sure you are correct Andras, and this is a cylindrical probe. From my research it seems the most popular type for that machine. I don’t have direct access to the machine, but I might be able to get some data for you if that would be of help. Thank you for all you do.

OK, thanks for the note. I’ll find some time to work on this soon.

Hi! thanks for your awesome work!! do you need anymore data??? i have a Voluson P8, if you need some volumetric data tell me