Motion Detection and Correction for Frame-Based Stereotactic Localization

Abstract

Frame-based stereotactic localization is an important step for targeting during a surgical procedure. The motion may cause artifacts in this step reducing the accuracy of surgical targeting. While modeling of motion in real-life scenarios may be difficult, herein we analyzed the case where motion was suspected to impact the localization step. In this case, a scan with and without motion was performed with a 3N localizer, allowing for a thorough analysis. Pseudo-bending of straight rods was seen when analyzing the data. This pseudo-bending appears to occur because head-frame motion during imaging acquisition decreases the accuracy of the subsequent reconstruction, which depends on Digital Imaging and Communications in Medicine (DICOM) metadata to specify the slice-to-slice location that assumes embedded object stability. Comparison of single-slice and multi-slice stereotactic localization allowed for comparative errors for each slice in a volume. This comparative error demonstrated low error when the patient was under general anesthesia and presumed not to have moved, whereas a higher error was present during the scan with motion. Pseudo-bending can be corrected by using only localizer fiducial-based information to reorient the pixels in the volume, thus creating a reoriented localizer scan. Finally, targeting demonstrated a low error of 0.1 mm (+/- 0.1 mm) using this reoriented localizer scan, signifying that this method could be used to improve or recover from motion problems. Finally, it is concluded that stability and elimination of motion for all images utilized for stereotactic surgery are critical to ensure the best possible accuracy for the procedure.