Record an Ultrasound sweep by OpenIGTLink

Dear ImFusion team,

I have an Aixplorer US device, which is not supported as a plugin in ImFusion. I would like to calibrate it. I have tried the following:

  1. Stream US images and Tracking data to ImFusion Suite by OpenIGTLink.
  2. Import OpengIGTlink in ImFusion Suite and add an ImageStream and Tracking. They can be visualized in Suite correctly
  3. Select both the added ImageStream and TrackingStream, right click Ultrasound->Calibration Wizard. I can go through the whole process successully. But the calibration result always looked wrong. I used the 9-cones model, and in the cailbration results, all cones are in one row and are close to each other, instead of 3x3 cones.
  4. I also would like to record an US sweep based on the live tracking stream and the live image stream, but don’t know how to do it. When I select the 2 streams and right click them, there is no ‘record Ultrasound Sweep’ in the menu list.

I have tried multiple times, but still got the wrong results. So I am wondering if I could do the calibration like this? Could I also record an Ultrasound Sweep with these two streams from OpenIGTLink? If only image data and tracking matrix are contained in the OpenIGTLink message, does ImFusion Suite add data like timestamp automtically?

Thanks in advance for your help.
Best regards,

Hi Luohong,

First of all, using OpenIGTLink to stream both images and tracking data is perfectly fine. The calibration algorithm, recorders, etc. don’t really care how the data got into the Suite. What might be the problem here: the US sweep recorder doesn’t show up in the context menu because the image stream is not detected to be an Ultrasound image stream. Try first using the algorithm “Set Image Stream Modality” and set it to Ultrasound on the image stream. Then, the sweep recorder should open.

Can you provide an example of the cone acquisition from both directions? I’m not sure I fully understand what is happening.

By the way, if you have an image stream that is too shallow, so you can’t make out the cones well, you can use any other object in a water bath for the Calibration Wizard. The final step doesn’t have a shape prior, it just uses image-based similarity optimization to find a good calibration matrix. Just avoid symmetries in the object (e.g. a ball would be a poor choice) to prevent ambiguities.


1 Like

Hi Oliver,
Thanks a lot for your help! This is quite helpful. I will try your suggestions next week and provide updates later.