Variables affecting the accuracy of stereotactic localization using computerized tomography

• Published in: Journal of neurosurgery Vol. 79; no. 5; p. 667
• Main Authors: Bucholz, R D, Ho, H W, Rubin, J P
• Format: Journal Article
• Language: English
• Published: United States 01.11.1993
• Subjects: Brain - diagnostic imaging; Brain Diseases - diagnostic imaging; Stereotaxic Techniques - standards; Tomography, X-Ray Computed - methods; Tomography, X-Ray Computed - standards
• ISSN: 0022-3085
• DOI: 10.3171/jns.1993.79.5.0667

Stereotactic localization using computerized tomography (CT) is increasingly employed to guide neurosurgical procedures in crucial areas of the brain such as the brain stem. This technique allows the surgeon to resect a lesion in its entirety while sparing critical areas of the brain. Thus, the parameters used for scanning should be selected for maximum accuracy. While the small pixel size of CT scanners suggests a high degree of precision in localization, there have been few systematic studies of this accuracy. The authors have studied the amount of error in localization created by variables such as CT scan thickness, interscan spacing, size of lesion, and method of computation when using the Brown-Roberts-Wells (BRW) stereotactic system. Over 1000 CT scans were made of a phantom composed of spheres of differing diameter and location. The CT slice thickness was varied from 1.5 to 5.0 mm, and interscan spacing was varied from 0.5 to 3.0 mm. The coordinates of the center of the spheres were calculated independently using the laptop computer supplied with the unit and also by a stereotactic computer which automatically calculates the center of the fiducials. The actual BRW coordinates of the sphere center were obtained using the phantom base and were then compared to the computer-calculated coordinates to determine error in localization. Variables with a significant effect on error included the scan thickness, interscan spacing, and sphere size. The mean error decreased 23% as the scan thickness decreased from 5.0 to 1.5 mm and 45% as the interscan spacing decreased from 3.0 to 0.5 mm. Mean error was greatest for the smallest sphere sizes. The two computational methods did not differ in error. This study suggests that, for critical areas of the brain or for small lesions, a scan thickness of 1.5 mm and interscan spacing of 0.5 mm should be employed.