Is there an equivalence of non-invasive to invasive referenciation in computer-aided surgery?

Various navigation systems with non-invasive patient referenciation and registration methods have been developed in times of minimal-invasive and computer-aided surgery. However, hard data proving the equivalence of different referenciation systems are missing. The present study investigated invasive and non-invasive referenciation systems with regard to overall navigation accuracy as well as navigation accuracy at specific anatomic locations. Four skull models were individually fabricated with a 3D printer based on patient's CT data sets and fitted with an individual customized silicone skin. 26 titanium screws on defined anatomic locations served as target fiducials. Two non-invasive referenciation systems (headband and headset) were compared with the invasive skull fixed reference array. Registration was done with laser surface scan. The mean accuracy was calculated and the target registration error for eight anatomical locations was measured. Mean accuracy was 1.3 ± 0.12 mm for the skull fixed reference array, 1.44 ± 0.24 mm for the headset and 1.46 ± 0.15 mm for the headband referenciation (non-significant). Navigation accuracy of the invasive referenciation system was significantly superior to the accuracy of both non-invasive systems on the ethmoid sinus with respect to the selected anatomic locations. In the midface headband referenciation was statistically significantly worse than the invasive system. Invasive and non-invasive referenciation systems seem to be on par in terms of overall navigation accuracy, but not regarding specific anatomic locations. Therefore, invasive referenciation systems should be preferred in high precision surgery.