Sensitivity analysis for registration of vertebrae in ultrasound and computed tomography images

More than 250,000 lumbar spine operations are performed annually in the United States. Because of this large number of projected surgeries, improvements in surgical procedures can have a tremendous practical impact. We are developing a method to quickly and accurately register surface points extracted from ultrasound images of a vertebra to surfaces extracted from computed tomography (CT) images. identification of surface points in ultrasound images is a challenging task, since points must be extracted in real time and since large portions of the vertebral surface are invisible in the ultrasound image. Thus, important questions are raised: Can accurate registration of pre- and intra-operative images be achieved using only part of the vertebral surface, and if so, which regions are the most important? To answer these questions, we have performed an analysis of the sensitivity of the registration process to various surface patches. The vertebral surface of a selected lumbar vertebra was extracted from CT scans, and sets of points representing surface information that will eventually be extracted from ultrasound images were collected with an optically tracked probe. These points, which covered the entire posterior area of the vertebra, were grouped into twenty surface patch combinations to study the sensitivity of the registration algorithm to different patches. Our results show that some surface patches cause no degradation in registration accuracy while others lead to disastrous results. Generally, the best results are achieved with combinations that include the tip of the spinous process and one or two symmetric patches on either side. One combination that works consistently well is the combination of the spinous process with the central laminar regions. This finding is important because these areas of the vertebral surface can easily be visualized in ultrasound images obtained in the transverse plane.