Image guidance during highly conformal radiotherapy requires accurate geometric calibration of the moving components of the imager. Due to limited manufacturing accuracy and gravity-induced flex, an x-ray imager’s deviation from the nominal geometrical definition has to be corrected for. For this purpose a ball bearing phantom applicable for nine degrees of freedom (9-DOF) calibration of a novel cone-beam computed tomography (CBCT) scanner was designed and validated. In order to ensure accurate automated marker detection, as many uniformly distributed markers as possible should be used with a minimum projected inter-marker distance of 10 mm. Three different marker distributions on the phantom cylinder surface were simulated. First, a fixed number of markers are selected and their coordinates are randomly generated. Second, the quasi-random method is represented by setting a constraint on the marker distances in the projections. The third approach generates the ball coordinates helically based on the Golden ratio, ϕ. Projection images of the phantom incorporating the CBCT scanner’s geometry were simulated and analysed with respect to uniform distribution and intra-marker distance. Based on the evaluations a phantom prototype was manufactured and validated by a series of flexmap calibration measurements and analyses. The simulation with randomly distributed markers as well as the quasi-random approach showed an insufficient uniformity of the distribution over the detector area. The best compromise between uniform distribution and a high packing fraction of balls is provided by the Golden section approach. A prototype was manufactured accordingly. The phantom was validated for 9-DOF geometric calibrations of the CBCT scanner with independently moveable source and detector arms. A novel flexmap calibration phantom intended for 9-DOF was developed. The ball bearing distribution based on the Golden section was found to be highly advantageous. The phantom showed satisfying results for calibrations of the CBCT scanner and provides the basis for further flexmap correction and reconstruction developments.

In EBRT of prostate cancer, safety margins can be reduced by the means of interfractional correction following IGRT, hence reducing doses to organs at risk. An innovative method for further rectal dose reduction was successfully introduced by spacer materials injected between prostate and rectum. Aim of the study was the investigation of the spacer's influence on inter- and intrafractional movements of the prostate. In a prospective matched-pair comparison, eight patients were investigated up to now. In all of them, four gold marker fiducials were implanted perineally into the prostate under endorectal sonographic guidance. Four of the patients additionally received a rectal spacer consisting of 10 mL polyethylene glycol (PEG) hydrogel. Fiducials were used to determine translations and rotations of the prostate. During 7 field IMRT, we captured MV portal images on the fly and registered them by means of a panel flex correction procedure. Two-dimensional marker positions were automatically detected by means of a marker kernel convolution algorithm. Three-dimensional marker positions were reconstructed from two oblique projections. Daily analysis of first pair of MV images (gantry angles 220° and 265°) was done to determine interfractional movements. Intrafractional movements were derived from shifted marker positions in the last pair of MV portal images (gantry angles 95° and 140°). The time gap in between the two reconstructions was 4.4 min on average. Eight hundred eighty-four images gained during 221 fractions (supine, flat couch, knee support, comfortably full bladder, empty rectum, no intraprostatic marker migrations >2 mm of more than one marker) were analyzed: 95 fractions in the spacer group and 126 fractions in the control group, respectively. Interfractional 3D vector translations and L-R rotations were found to be 9.4 +/- 4.5 mm / 5.5° +/- 6.4° with spacer, 7.5 +/- 3.5 mm / 2.2° +/- 2.9° without spacer, respectively.(mean of means). Intrafractional movements were 1.9 +/- 2.1 mm / 2.3° +/- 2.7° in the spacer cohort, 1.9 +/- 1.1 mm / 2.0° +/- 2.8° in the control cohort. Differences were statistically not significant. Our findings compare to previously published 9.3 +/- 4.4 mm / 5.3° +/- 4.9° inter- and 3.0 +/- 3.7 mm / 2.5° +/- 2.3° intrafractional movements of 39 patients / 833(1013) fractions without spacer. Distensions of the anterior rectal wall could successfully be provided in all spacer patients. Spacers so far did not show a stabilizing effect on motion degree and direction between and during treatment fractions. Both cohorts will be completed to a sample size of 20 each. Whereas spacer application leads to a significant dose reduction in the anterior rectal wall, inter- and intrafractional mobility of the prostate was not affected, which has to be taken into account in safety margin design.