Artikel
Application accuracy of two neuronavigation systems: comparison and experimental study of registration methods
Suche in Medline nach
Autoren
-
D. Paraskevopoulos
- Neurochirurgische Klinik, Ruprecht-Karls Universität Heidelberg
-
R. Metzner
- Institut für Medizinische Biometrie und Informatik, Universität Heidelberg
-
M. Schrey
- Klinik für Neurochirurgie, Klinikum Nürnberg
-
J. Dreyhaupt
- Institut für Medizinische Biometrie und Informatik, Universität Heidelberg
-
A. Unterberg
- Neurochirurgische Klinik, Ruprecht-Karls Universität Heidelberg
-
C. Wirtz
- Neurochirurgische Klinik der Universität Ulm am Bezirkskrankenhaus Günzburg
| Veröffentlicht: | 20. Mai 2009 |
|---|
Gliederung
Text
Objective: The study aimed at testing and comparing accuracy of an active and a passive optical neuronavigation system. Furthermore, surface registration techniques, often resulting into decreased application accuracy, were to be experimentally studied.
Methods: A Stryker Leibinger and a BrainLAB VectorVision Sky navigation system were tested. CT and MR series were acquired and two different types of phantoms used, one for accuracy testing and a non rigid one specifically designed for surface registration techniques. In order to calculate software and system accuracy respectively, true coordinates of target points were acquired by a high precision Computer Numerical Control (CNC) machine and compared to coordinates measured by the systems in image data sets and navigation procedures. Comparison of image data set coordinates and navigation coordinates was used to evaluate navigation accuracy. Statistics used were regression models and ANOVA.
Results: The systems showed similar results, achieving overall navigation accuracies of less than 1.5 mm. Stryker had a significantly better software accuracy, while BrainLAB demonstrated significantly better system and navigation accuracy (p<0.01); nevertheless, differences were submillimetric. Precision data from the systems did not reflect the measured accuracy.
Surface matching resulted in no significant improvement of accuracy, confirming former studies. Mean Euclidean deviation of both systems after surface matching were on the same level. Additional surface matching with the Stryker system improved inaccurate fiducial registrations in a park-bench position for posterior fossa targets. Total number and accepted percentage of acquired surface points were significant factors. Laser registration with BrainLAB Z-touch showed no significant differences compared to conventional pointers.
Conclusions: Passive and active optical navigation systems have reached similar levels of accuracy. Slight differences may prove statistically significant but are limited as absolute values, their clinical relevance thus being restricted. Surface registration may improve inaccurate fiducial registrations under certain circumstances; however it does not generally improve accuracy.
