Article
Accuracy measurement of Volume-CT using a phantom
Search Medline for
Authors
Published: | January 27, 2009 |
---|
Outline
Text
Introduction
One of the limiting factors of current navigated surgery is the limited resolution of state-of-the-art imaging modalities such as MSCT. New CT scanner designs based on high resolution flat-panel detectors could overcome these limitations. This new scanner design, widely known as flat-panel based Volume-CT (fpVCT, see Figure 1 [Fig. 1] on left side), makes it possible to acquire volume datasets at a Nyquists resolution limit of 200 µm, while current CT allows merely 400–500 µm. This improvement in spatial resolution of high contrast structures may improve the accuracy of intraoperative navigation.
Materials and methods
To investigate this, we prepared VCT datasets of a calibration phantom with a fpVCT prototype system (GE CR&D®, Schenectady, NY, USA) and a current MSCT system (Lightspeed 16, GE Healthcare®, see Figure 1 [Fig. 1] on right side) using a resolution optimized scan protocol. Imaging data were transferred to the passive optoelectronic VectorVision² navigation system (BrainLAB®, Heimstetten, Germany). The phantom was marked with diverse crater-shaped pits and osteosynthesis screws, which were arranged in all directions in space. A reference adapter was tightly attached to the phantom. After manual identification of the markers within the 3D data record, they were localized and registered on the phantom under navigation surveillance with a straight pointer. 3D measurements were performed to check the target registration error (TRE).
Results
Average TRE using MSCT was 0.66 (standard deviation 0.32) mm and using fpVCT 0.41 (standard deviation 0.13) mm. The distribution of all TREs was positively checked for normalization using a Kolmogorov-Smirnov (Lilliefors) test of normality. TRE differences of MSCT and VCT were statistically significant to a level of p=0.0002 (see Figure 1 [Fig. 1] in the middle).
Discussion
One has to note that the performance of navigation technology is a complex problem. Nevertheless, the navigation cannot be better than the underlying imaging: Standard clinical MSCT imaging protocols provide a voxel size of about 0.5 mm³. Current navigation surgery delivers orientation to the surgeon with an accuracy of about 1 to 1.5 mm depending on the methods of registration and referencing used. Current accuracy is sufficient for navigated interventions of the frontal skull base e.g. sinus surgery. Surgery of the lateral skull base with its exceptionally difficult anatomy demands better navigation precision. For example, Schipper et al. identified a TRE <0.5 mm as the required accuracy for performing a navigated cochleostomy. The VCT imaging seems to fulfill these needs and to open new possibilities for navigated surgical procedures e.g. by use of Navigation Control® or intraoperative mechatronic assistance systems. It might be, that after increasing the underlying imaging dataset’s accuracy the inherent accuracy of the navigation system could play a much greater role then. Further clinical studies on this issue have to be carried out to bring the new imaging technology into practice and to evaluate wether the navigation precision can be achieved in a clinical setting too.