Quick Links

Home

Documentation

Download

Availability

---

News:

Studierstube 4.4
released

---

 

Cranial Ultrasound Simulation System

CranUS is an augmented reality based simulator for Neonatal Cranial Sonography. Its purpose is to supply medicals with training and education at this special Sonography method at objects which - opposited to real infants - do not move around, which is the perfect training situation.

The cause of using ultrasound for neonatal examination is simple - it does not cause damage to the tissue which is examined like XRay does, it is not invasive, and does not require the patient to keep still as much as at for example CT.
Cranial sonography is only possible in the first year of a child because the fontanelles of the bone plates of the head are still soft tissue, which means that ultrasound can penetrate it. So medicals can examine the head of the baby regularly and are able to detect possible abnormalities.

For being able to provide realtime rendered slices through a cranial volume, this volume was created by resampling 2d-ultrasound slices to a 3d-texture of a brain. After some postprocessing (interpolation algorithms for reducing artifacts etc.) the volume is ready for use in the Application.

The training environment itself is provided with an ART-infrared-tracking system and different props to be tracked, in this case a phantom (a anatomical correct baby doll) and a real ultrasound probe. These two props are augmented with rethrow-reflecting markers which can be tracked by the infrared cameras. It is recommended that these markers are mounted in a very robust way so if one of them is missfortunately dropped, it does not take damage and is not deformed which can cause issues with the tracking system.

To use the application as an augmented reality based training system, it was necessary to create virtual models of the props which are tracked. The construction of the model of the probe was done with a 3d model generation tool while the model of the phantom was constructed with CT pictures taken from it. The correct parts of the doll were labelled in the pictures and with this created labelmapt a volume was constructed.

The software itself contains two main modules.
First, there is an application to construct a 3d cranial Volume from recorded ultrasound pictures and applies some algorithms for interpolations and artifact removing.
The second part is the training application itself. The main window consists of four frames.

The upper right frame shows the virtualized models of the phantom and the probe and is provided with tracking data from the ART system. It shows the two props' movements in real time and is also used for calibrating the position and orientation of the cranial 3d volume which is connected to the phantom's movements and should be positioned somewhere in the head of the virtual phantom.

The upper left frame is used for displaying the picture a real ultrasound probe would record when it would be moved over an infant's head. The rendering of this picture is done as follows:

• First, the current position of the cranial volume is derived, dependent on the position of the phantom.
• Next, a cutting plane is built. It is connected to the virtual probe and its movements
• The application now calculates the relative transform (rotation, translation) between the volume and the cutting plane and if the plane cuts through the volume, the 3d texture of the volume is mapped to a 2d texture on a plane and is displayed in grayscale.
• For providing better training results, a mechanism which calculates if the probe is held approximately to the correct position (the fontanelle) of the baby's head decides if the display stays black (too far away from the fontanelle) or if the volume is rendered on the 2d plane

The lower left frame is used for displaying some information.
The lower right frame shows the tracking data of the phantom and the probe and visualizes the update rate.