Image-Guided Neurosurgery
Graduate Research Students: Kevin & Jessie
Frame-based stereotaxy is a method which uses 3D coordinates to locate a target in the area of the brain with millimetric precision. The coordinates may be obtained by fitting a patient's head with a special frame and taking a CT or MRI scan. The position of the brain structures relative to the frame permits fine localization of the deep brain structures.
Accurate frame-based stereotaxy depends on:
- Positioning - The position of the frame must be precise.
- Reducing Spatial Errors - Minimizing this type of error depends on the type of imaging modality used. We are currently investigating the use of MRI versus ultrasound.
- Frame Properties - The frame itself has mechnanical properties that influences accuracy.
- Target computation - Algorithms used for target computation rely on the accurate selection of fiducial points in the image. Errors in the selection of these points will propagate through the computation to degrade the final result. This is the largest source of application error.
Our Contribution
In our laboratory, we present an target computation algorithm that can automatically identify the fiducial markers. A transformation is computed based on the volumetric configuration of the fiducial marker set, rather than on individual slices through the markers. This algorithm is (currently?) undergoing validation with a phantom study. We compare the computationally derived target locations with locations found mechanically using a Leksell G frame and arc.
This technique has a particular application to MR where the algorithm's automatic monitoring of image acquisition quality minimizes errors.
Using ultrasound guidance
Because the brain shifts position when craniotomy procedure is performed, image-guided surgery systems that rely on a pre-operative MR image are insufficient for a broad range of neurosurgical tasks. We are currently developing techniques that use ultrasound images of the brain, collected during the surgery, to track the distortion of the brain tissue. Once the distortion has been measured, we can use it to warp the pre-operative MR images. This helps to accurately reflect the shape of the brain during surgery.
How are 3D ultrasound image volumes constructed?
To create 3D ultrasound image volumes, the video from the tracked ultrasound probe is reconstructed using a real-time reconstruction technique. The ultrasound volume is built up step-by-step as the video frames are acquired. The ultrasound can be visualized through volume rendering techniques, as shown below, or painted over the three MR image planes. The images below are set up for stereo viewing
.
MRI vs. Ultrasound Video
We compare the ultrasound images to an MR image volume that was acquired prior to surgery. In order for this to be done, the ultrasound probe must be tracked in 3D space to monitor it's position relative to the MR image coordinate system. We use the POLARIS (shown below) an infrared tracking system from Northern Digital.
The image on the left an overlay of the ultrasound video (right) over the pre-operative MRI (left).
The MR image is re-formatted in real time as the ultrasound operator moves the probe. The image above displays the orientation of the ultrasound slice relative to the canonical axes of the MR volume.
Because the overlay is done in real time (10 frames per second), the surgeon is able to very rapidly obtain a qualitative assessment of the shift. If the shift is not significant, then the surgeon has the option to immediately continue the surgery without any shift correction.
