| Autor: |
Goos, Gerhard, Hartmanis, Juris, Leeuwen, Jan, Wells, William M., Colchester, Alan, Delp, Scott, Li, Qinghang, Zamorano, Lucia, Jiang, Zhaowei, Vinas, Fernando, Diaz, Fernando |
| Zdroj: |
Medical Image Computing & Computer-Assisted Interventation - MICCAI'98; 1998, p253-260, 8p |
| Abstrakt: |
The purpose of this study was to determine the application accuracy of a new frameless marker system for interactive intraoperative localization of intracranial lesions. The influence of image quality, registration error, repeatability, and marker distribution on the application accuracy were analyzed and compared. A phantom was mounted with the standard Z-D ring and also implanted with frameless marker system, which randomly distributed on the surface. The phantom was scanned as routine with 1 mm and 2 mm sections. The pixel sizes were used 1.18×1.18 and 0.59×0.59. The two systems were tested under different image quality and registration. The target point was digitized and the coordinates were recorded and compared with reference points. The difference between two systems were tested with paired t-test. Image data were loaded into a SUN Workstation and registered with NSPS.4.0 software. The coordinate of each fiducial marker was recorded into a file as the reference. The tip of each semi-invasive fiducial marker was digitized to achieve a frameless transformation matrix, and the special points on the Z-D ring were digitized to achieve a frame-based transformation matrix. The differences from the reference points were used as the deviation from "true point". The mean square root (RMS) was calculated to show the sum of vectors. The results of 2 mm section group showed that the registration error of frame-based system is 3.42±0.22 mm and the error of the frameless system is 1.01±0.63 mm (P<0.001). The RMS are 2.57±0.54 mm and 1.53±0.65 mm respectively (P<0.001). The RMS of error registration (one point off 5 mm) are 5.01±0.26 mm and 2.23±0.13 mm respective (P=0.003). The results of 1mm section group showed that the RMS are 1.20±0.42 mm and 0.90±0.47 mm respectively (P=0.121). The higher the quality (the thinner scan thickness) of image it is, the better the application accuracy will be (P=0.001 and 0.032 respectively). These preliminary results showed that the frameless semi-invasive fiducial marker system can provide clinical acceptable accurate localization as the frame based surgical localization system did. There is no significant difference between the experimental and clinical results. The higher the quality of image it is, the better the application accuracy will be. But there is no significant difference between 1mm sections and 3 mm sections of MRI images. [ABSTRACT FROM AUTHOR] |
| Databáze: |
Supplemental Index |
| Externí odkaz: |
|
