Integral vidcography overlay navigation system using mutual information-based registration

Abstract

We propose an automatic markerless registration method for Integral Videography (IV) overlay navigation system that directly matches a video image of the patient to a 2-D rendering of a surface model extracted from the 3-D surface model image. IV overlay system is integrated into a surgical navigation system to superimpose a real three-dimensional (3-D) autostereoscopic image onto the patient via a half-silvered mirror. This registration technique doesn't require the setting of markers, which is invasive and extends the duration of the procedure. Accuracy measurements were performed using a triangular-prism- shaped model. The mean Target Registration Error (TRE) was 1.6 ± 0.6 mm. We also applied our method to a human face registration and obtained successful results, although some noises were added to the video image. The presented method is more suitable for clinical applications of IV image overlay than previous methods since the presented method is sufficiently accurate, robust and more convenient for surgeon. © Springer-Verlag 2004.