Session Item

To analyze the effect of intra-fraction patient shifts in frameless LINAC-based stereotactic radiosurgery (SRS) of cranial metastases and how to handle it. In detail, we investigate if the use of intra-fractional image-guidance and 6 degree-of-freedom (6D) positioning correction is necessary for single-isocenter SRS with several couch angles, especially when multiple lesions are targeted. 

Targeting accuracy is measured on a Varian TrueBeam STx linac using the BrainLab 6D-ExacTrac system. For immobilization a standard stereotactic head mask (BrainLab) was utilized. Intra-fractional patient shift data is obtained after each couch angle variation (couch kick) by positioning verification with a pair of stereoscopic x-ray images auto-matched to planning CT data. When measured shifts are within tolerance (0.5 mm translational and 0.5° rotational), the treatment is continued. Otherwise, 6D-corrections will be applied and additional x-ray images have to be acquired to verify whether patient position is within tolerance again. The resulting 6D-shifts measured directly after couch kick (no corrections) will be compared with the finally measured (tolerable) 6D-shifts.

121 SRS cases (85 single and 36 multiple target) with 280 couch kicks in total were evaluated. For 66.8 % of couch kicks the specified tolerances were exceeded and patient position had to be corrected.

Intra-fractional translational errors (one standard deviation [SD]) were 0.50 vs. 0.22 mm in left-right, 0.46 vs. 0.22 mm in superior-inferior, and 0.37 vs. 0.21 mm in anterior-posterior direction (see figure), with a mean three-dimensional vector of 0.71 vs. 0.33 mm [without versus with positioning readjustment after couch kick].

Furthermore, rotational errors were 0.43 vs. 0.28° in pitch, 0.37 vs. 0.25° in roll, and 0.34 vs. 0.19° in yaw rotation.

Finally, the change in dose distribution due to uncorrected 6D errors was calculated to assess the effect on target coverage.

A few single target (9 %) and almost all single-isocenter multiple target cases (91 %) would be clinically unacceptable (coverage loss >5 %) if the positioning readjustment had not been applied. The mean lesion size was 1 cc (SD 0.9 cc) and the average PTV-center to isocenter distance for multiple targets was 4.8 cm (SD 1.7 cm).

Scatter plots with mean and 95% confidence regions of translational positioning shifts for all 280 couch kicks: uncorrected (a-c) and corrected (d-f) resulting in tolerable deviations (0.5 mm, 0.5°)

We conclude that initial setup verification is not enough for frameless LINAC-based SRS and may lead to considerable positioning error due to intra-fractional patient motion and geometrical uncertainties unless this is monitored and corrected for during treatment.

Especially for single-isocenter treatments of multiple targets, the use of a 6D positioning system is highly recommended in order to minimize unfavorable off-isocenter displacements leading to insufficient coverage and treatment outcome.