An algorithm to auto-match the fiducial markers for CT Image-Guided Surgical Navigation System

The performance of matching the fiducial markers plays an important role in the accuracy of the CT Image-Guided Surgical Navigation System (IGNS). This paper proposed a new algorithm for finding markers in CT image automatically and realizing the one-to-one correspondence between the fiducial markers of CT Image space in IGNS and the sensors of Cartesian space in magnetic field. Firstly, the CT images are segmented by the 3D region growing method. Secondly, upon criteria which is using the length and volumn of the sensors, several suspicious connected components are identified as the possible markers. After the identification, the 3D center of the suspicious connected components are computed. In the end, the one-to-one correspondence is made by calculating the correlation coefficients between the coordinates of suspicious connected components in image space and that of the sensors in magnetic field. The algorithm was applied to both the adaptability and the precision experiments on two models. The result shows that even the precise parameters of the markers are not known, most of them can still be found quickly. And the errors reach the sub-voxel precision. The algorithm is stable, self-adaptive and easy to use.