SU-E-J-205: Modeling Fiducial Marker Movement in Prostate

To understand fiducial marker movement in prostate and to improve prostate patient positioning with fiducial markers.

Patient planning CT and positioning CBCT were retrospectively analyzed. CBCT images were automatically registered to the corresponding CT ones based on their bony structures in Eclipse. The coordinates of the markers in the CT and CBCT were recorded. The status of bladder and rectum fillings, and gas in rectum were indexed into four levels and recorded. An algorithm was developed to optimally register the (three) fiducial markers in CBCT with those in the CT so that only the information on prostate deformation and marker migration remained. A mathematical model was built to estimate the marker migration and the relative contribution to the deformation from the status change of bladder and rectum.

The registration error from our optimal registration algorithm indicates the minimal patient setup error by using fiducial markers. The results from our analyzed 3 patients were average(standard deviation) = 1.10(0.40), 1.27(0.67) and 0.62(0.38) mm, but there were some fractions when the registration errors were over 2 mm. From our mathematical model, it was found that the marker migration could be as large as 2.9 mm. If 2 mm tolerance is required in such as SBRT, then soft tissue based registration may be more appreciated when the optimal registration error is greater than 2 mm or from the date when the migration is greater than 2 mm because the migrated marker may not change the location thereafter. In addition, for one investigated patient, the relative contribution to the deformation from the bladder, rectum filling and gas in the rectum is 0.43:0.31:0.26. More patient data are in analysis, and other interesting results will be reported.

Fiducial marker migration can be over 2 mm. The migration will Result in a conclusion that small registration error does not mean good patient positioning.