Technical note: A method for generating lesion-specific nonuniform rotational margins for targets remote from isocenter.

Abstract

To present a novel method for generating nonuniform lesion-specific rotational margins for targets remote from isocenter, as encountered in single isocenter multiple metastasis radiotherapy. Target contours are rotated using a large series of 3D rotations, corresponding to a given range of rotational uncertainty, and combined to create a rotational envelope that encompasses potential motion. A set of artificial spherical targets ranging from 0.5 to 2.0 cm in diameter, and residing a distance of 1 - 15 cm from isocenter, is used to generate rotational envelopes assuming uncertainties of 0.5-3.0°. Computing time and number of samples are reported for simulated scenarios. Hausdorff distances (HD) between rotational envelopes and original target structures are calculated to represent the magnitude of uniform expansion required to encompass potential rotation. Volume differences between uniform expansions (based on HD) and rotational envelopes are reported to articulate potential advantages. Median time to generate rotational envelopes was 60 s (31-974 s). Median required samples was 86 (61-851). Maximum HD for all targets located 10 cm from isocenter was 1.5 mm, 3.0 mm, 5.8 mm, and 8.6 mm assuming 0.5°, 1.0°, 2.0°, and 3.0° of rotational uncertainty, respectively. At 5 cm from isocenter and assuming 0.5° of rotational uncertainty, volumes were decreased by 0.07 cc (60%), 0.24 cc (39%), and 1.08 cc (19%) for 5 mm, 10 mm, and 20 mm targets respectively. At 10 cm from isocenter and 1.0° of uncertainty, volumes decreased by 0.42 cc (58%), 2.0 cc (40%), and 2.5 cc (27%). On average target volumes decreased 45% (SD = 17%) when compared with uniform expansions based on HD. Rotational margins may be generated by sampling a set of 3D rotations. Resulting margins explicitly account for target shape, distance from isocenter, and magnitude of rotational uncertainty, while reducing treated volumes when compared with uniform expansions.