Registration accuracy and possible migration of internal fiducial gold marker implanted in prostate and liver treated with real-time tumor-tracking radiation therapy (RTRT).

Background And Purpose: We have developed a linear accelerator synchronized with a fluoroscopic real-time tumor-tracking system to reduce errors due to setup and organ motion. In the real-time tumor-tracking radiation therapy (RTRT) system, the accuracy of tumor tracking depends on the registration of the marker's coordinates. The registration accuracy and possible migration of the internal fiducial gold marker implanted into prostate and liver was investigated.

Materials And Methods: Internal fiducial gold markers were implanted in 14 patients with prostate cancer and four patients with liver tumors. Computed tomography (CT) was carried out as a part of treatment planning in the 18 patients. A total of 72 follow-up CT scans were taken. We calculated the relative relationship between the coordinates of the center of mass (CM) of the organs and those of the marker. The discrepancy in the CM coordinates during a follow-up CT compared to those recorded during the planning CT was used to study possible marker migration.

Results: The standard deviation (SD) of interobserver variations in the CM coordinates was within 2.0 and 0.4 mm for the organ and the marker, respectively, in seven observers. Assuming that organs do not shrink, grow, or rotate, the maximum SD of migration error in each direction was estimated to be less than 2.5 and 2.0 mm for liver and prostate, respectively. There was no correlation between the marker position and the time after implantation.

Conclusion: The degree of possible migration of the internal fiducial marker was within the limits of accuracy of the CT measurement. Most of the marker movement can be attributed to the measurement uncertainty, which also influences registration in actual treatment planning. Thus, even with the gold marker and RTRT system, a planning target volume margin should be used to account for registration uncertainty.