Comparison of different registration methods for surgical navigation in cranio-maxillofacial surgery

Abstract

Surgical navigation requires registration of the pre-operative image dataset with the patient in the operation theatre. Various marker and marker-free registration techniques are available, each bearing an individual level of precision and clinical practicability. In this study the precision of four different registration methods in a maxillofacial surgical setting is analyzed. A synthetic full size human skull model was registered with its computer tomography-dataset using (a) a dentally mounted occlusal splint, (b) the laser surface scanning, (c) five facial bone implants and (d) a combination of dental splint and two orbital bone implants. The target registration error was computed for 170 landmarks spread over the entire viscero- and neurocranium in 10 repeats using the VectorVision2 (BrainLAB AG, Heimstetten, Germany) navigation system. Statistical and graphical analyses were performed by anatomical region. An average precision of 1mm was found for the periorbital region irrespective of registration method (range 0.6-1.1mm). Beyond the mid-face, precision linearly decreases with the distance from the reference markers. The combination of splint with two orbital bone markers significantly improved precision from 1.3 to 0.8mm (p<0.001) on the viscerocranium and 2.3-1.2mm (p<0.001) on the neurocranium. An occlusal splint alone yields poor precision for navigation beyond the mid-face. The precision can be increased by combining an occlusal splint with just two bone implants inserted percutaneously on the lateral orbital rim of each side.