| Three-dimensional ultrasound (3-D US) is finding a host of applications among the many surgical fields. Tracked, freehand 3-D US can be utilized in spinal surgery as a means to identify vertebral regions of interest and guide the surgeon towards the target prior to any incision being made. If 3-D US images are used to establish a transformation between the patient space in the OR and image space, other tracked instrumentation can then be introduced to further guide the surgeon during the procedure.; In this work, an optically-tracked device for acquiring 3-D US information was designed, constructed and calibrated to an accuracy of approximately 0.4 mm. The device subsequently was tested in an application for image-guided surgery of the spine. Using a surface-based image registration algorithm, US images were mapped to pre-operative CT images from the lumbar vertebrae of phantoms and humans.; This method of registration first requires segmentation of the vertebral regions of interest. The vertebral surfaces from CT and US images were extracted using an intensity-based method. A rigid image registration was performed using an independent implementation of the Iterative Closest Point (ICP) algorithm. Accuracy measures demonstrat ed that the vertebral surfaces can be aligned to millimetric levels of accuracy. When an accurate registration was achieved, the resultant transformation was able to predict the trajectory of a tracked tap for pedicle screw insertion. Results further indicated that the accuracy of registration is dependent upon the frequency of the US signal, the resolution of the CT images, and the intensity at which the surfaces are extracted; Prior to clinical use, future work is needed to determine the best combination of these parameters. This approach to image-guided spinal surgery may also be beneficial for other regions of the spine, such as the thoracic region, where obtaining clear fluoroscopic images for guidance is hindered by the anatomy. This study is the first of its kind, accurately mapping 3-D vertebral US to CT images with an independent measure of registration accuracy... |
