Voxel-based 2D/3D registration enables the use of preoperative 3D CT scans during X-ray fluoroscopy guided interventions for navigation purposes. Both high accuracy and short computation times are crucial. The registration method is described. Most time consuming is the computation of a perspective projection from the 3D data. This paper compares several volume visualisation methods to compute such a projection quickly and accurately. The advantages and disadvantages of the methods in regard to our problem are discussed. Direct volume rendering methods proved to be particularly suitable and two specific methods were evaluated using a clinical dataset. Several optimisations of these methods in regard to 2D/3D registration are discussed and the results of two appropriate optimisations on the test dataset are shown. A method based on a shear-warp factorisation of the viewing transformation combined with runlength encoded data showed the best results.
|Title of host publication||First International Workshop on Biomedical Image Registration|
|Subtitle of host publication||(WBIR 1999)|
|Editors||Franjo Pernus, Stanislav Kovacic, H. Siegfried Stiehl, Max A. Viergever|
|Place of Publication||Ljubljana|
|Publisher||Slovenian Pattern Recognition Society|
|Number of pages||14|
|Publication status||Published - 30 Aug 1999|
|Event||International Workshop on Biomedical Image Registration WBIR’99 - Bled, Slovenia|
Duration: 30 Aug 1999 → 31 Aug 1999
|Conference||International Workshop on Biomedical Image Registration WBIR’99|
|Period||30/08/99 → 31/08/99|
GOECKE, R., Weese, J., & Schumann, H. (1999). Fast Volume Rendering Methods for Voxel-based 2D/3D Registration – A Comparative Study. In F. Pernus, S. Kovacic, H. S. Stiehl, & M. A. Viergever (Eds.), First International Workshop on Biomedical Image Registration: (WBIR 1999) (pp. 89-102). Ljubljana: Slovenian Pattern Recognition Society.