Loading GE Ultrasound

Support
,
1

Hi all,

I have a set of US data from a couple of patients that were exported from EchoPAC with the idea of working with raw data to train an AI model. The folder structure is:

TEST02
---DICOMDIR
---GEMS_DB
      |____ArcAudit.dbs
      |____ArcData.dbs
      |____ArcParam.dbs
      |____Master.db
      |____Master.log
---GEMS_IMG
      |____---2019_DEC
               |____---16
                        |____---JV123120
                                 |____JCGCLG3C
                                 |____JCGCLK3E
                                 |____ ...

I installed the SlicerHeart extension and followed the instructions in SlicerHeart/Docs/ImageImportExport.md at master · SlicerHeart/SlicerHeart · GitHub (I got the DLL file and run regsvr32 Image3dLoaderGe.dll as administrator). Then I load the data from the DICOM database using the DICOMDIR file. As a result only two volumes (out of 68 files) get loaded as 3D volumes with time, each with an ECG table as shown in the image.

image
image1915×992 153 KB

The remaining files fail to load as US, but I can drop them and load as volumes. Though in that case they look like screen shots as shown below.

image
image1918×993 205 KB

For the other patient data, which has the same structure, when loading with the DICOMDIR I get a warning message:

image
image526×206 11.9 KB

and when loaded the shape of the image is rectangular and no table appears, though the frame sequence runs with time.

image
image1918×993 161 KB

How can I load all the data correctly? or is there a problem with the data? In that case, anyone knows how to export it to get all the infor and correct formats?

Operating system: Windows 11
Slicer version: 5.6.2

2

Loading of 3D+t ultrasound images look good.

You also seem to have some 2D+t ultrasound acquisitions, which are essentially screen captures from the ultrasound system. They are usually littered with burnt-in annotations and they are only 2D, but you can still make 2D measurements on them.

We improved import of these images in recent Slicer Preview Releases, so I would recommend to use that instead of the latest Slicer Stable Release (currently Slicer-5.6.2).

You may also find the Ultrasound extension useful. It can help with post-processing and annotating ultrasound image sequences for deep learning training.