| dc.description.abstract | The procedures of spinal decompression, fixation, and fusion are moving towards
minimally invasive surgery. The need to repeatedly take X-ray images to confirm the
positions of surgical instruments and the need for surgeons rich clinical experience has
led to the development of surgical navigation systems. Among many navigation systems,
the equipment cost of the navigation system based on 2D C-arm image is lower, but the
2D C-arm image information to provide path planning is insufficient. The navigation
system based on 3D C-arm or O-arm image provides 3D image information, but the
equipment costs higher. These navigation systems that only use intraoperative C-arm
images do not integrate MR images that contain a large amount of soft tissue information
for physicians to use for diagnosis and precise surgical path planning.
This research integrates the 2D C-arm image navigation system with CT images to
achieve the advantages of low cost of equipment and 3D image path planning. In addition,
the registration of MR and CT images enables the surgeon to plan more precise
endoscopic path for spinal decompression and screw insertion path for spinal fusion than
that provided only by CT images, so that it will reduce the operation time and unexpected
trouble or mistakes during the operation. In this study, the curved surface mapping method
is used as the main registration method for MR and CT images. Use algorithm to search
for spinal cord area in MR images and generate its three-dimensional model, and to find
the contour points of the spinous process and the upper edge of the vertebral foramina as
the registration points for the registration of the bone contour surface reconstructed from
the CT images. The registration between MRI and CT images uses the iterative closest
point method with initial feature points registration to speed up the convergence
effectively. Intraoperative alignment of CT and C-arm images is based on the way
surgeons normally take C-arm images. The DRR images similar to C-arm images are
generated from CT images, and auxiliary line for comparison and calibration is also
provided for surgeons to plan on CT images. The surgical path on the DRR (Digitally
Reconstructed Radiography) is therefore easily mapped to the C-arm image. In addition,
an endoscope holder for decompression surgery has also been developed. It is easy to
adjust the angular direction of the endoscope about a fixed point in space and hold the
endoscope, so that the surgeon does not need to hold the endoscope manually.
The accuracy of MR and CT image registration was evaluated experimentally using
marker-attached prostheses. The experimental results show that if the displacement or
angular errors of initial alignment are within ±4mm or within ±1 degree respectively, the
registration using iterative closest point method will converge with an error within 2mm,
which satisfies the clinic need of spine surgery. In addition, the DRR image generated by
the CT image is compared with the C-arm image. The cross auxiliary line can provide an
effective comparison to help the surgeon adjust the X-ray projection position and angle,
so as to the planned path prior to the operation can be mapped to the C-arm image and
used by the navigation system for guidance and positioning. | en_US |