Motion monitoring for cranial frameless stereotactic radiosurgery using video-based three-dimensional optical surface imaging.

PURPOSE: To establish a new clinical procedure in frameless stereotactic radiosurgery (SRS) for patient setup verification at treatment couch angles as well as for head-motion monitoring during treatment using video-based optical surface imaging (OSI). METHODS: A video-based three-dimensional (3D) OSI system with three ceiling-mounted camera pods was employed to verify setup at treatment couch angles as well as to monitor head motion during treatment. A noninvasive head immobilization device was utilized, which includes an alpha head mold and a dental mouthpiece with vacuum suction; both were locked to the treatment couch. Cone beam computed tomography (CBCT) was used as the standard for image-guided setup. Orthogonal 2D-kV imaging was applied for setup verification before treatment, between couch rotations, and after treatment at zero couch angle. At various treatment couch angles, OSI setup verification was performed, relative to initial OSI setup verification at zero couch angle after CBCT setup through a coordinate transformation. For motion monitoring, the setup uncertainty was decoupled by taking an on-site surface image as new reference to detect motion-induced misalignment in near real-time (1-2 frames per second). Initial thermal instability baseline of the real-time monitoring was corrected. An anthropomorphous head phantom and a 1D positioning platform were used to assess the OSI accuracy in motion detection in longitudinal and lateral directions. Two hypofractionated (9 Gy x 3 and 6 Gy x 5) frameless stereotactic radiotherapy (SRT) patients as well as two single-fraction (21 and 18 Gy) frameless SRS patients were treated using this frameless procedure. For comparison, 11 conventional frame-based SRS patients were monitored using the OSI to serve as clinical standards. Multiple noncoplanar conformal beams were used for planning both frameless and frame-based SRS with a micromultileaf collimator. RESULTS: The accuracy of the OSI in 1D motion detection was found to be 0.1 mm with uncertainty of +/- 0.1 mm using the head phantom. The OSI registration against simulation computed tomography (CT) external contour was found to be dependent on the CT skin definition with -0.4 mm variation. For frame-based SRS patients, head-motion magnitude was detected to be <1.0 mm (0.3 +/- 0.2 mm) and <1.0 degree (0.2 degrees +/- 0.2 degrees) for 98% of treatment time, with exception of one patient with head rotation <1.5 degrees for 98% of the time. For frameless SRT/SRS patients, similar motion magnitudes were observed with an average of 0.3 +/- 0.2 mm and 0.2 degrees +/- 0.1 degree in ten treatments. For 98% of the time, the motion magnitude was <1.1 mm and 1.0 degree. Complex head-motion patterns within 1.0 mm were observed for frameless SRT/SRS patients. The OSI setup verification at treatment couch angles was found to be within 1.0 mm. CONCLUSIONS: The OSI system is capable of detecting 0.1 +/- 0.1 mm 1D spatial displacement of a phantom in near real time and useful in head-motion monitoring. This new frameless SRS procedure using the mask-less head-fixation system provides immobilization similar to that of conventional frame-based SRS. Head-motion monitoring using near-real-time surface imaging provides adequate accuracy and is necessary for frameless SRS in case of unexpected head motion that exceeds a set tolerance.

Pencil beam approach for correcting the energy dependence artifact in film dosimetry for IMRT verification.

The higher sensitivity to low-energy scattered photons of radiographic film compared to water can lead to significant dosimetric error when the beam quality varies significantly within a field. Correcting for this artifact will provide greater accuracy for intensity modulated radiation therapy (IMRT) verification dosimetry. A procedure is developed for correction of the film energy-dependent response by creating a pencil beam kernel within our treatment planning system to model the film response specifically. Film kernels are obtained from EGSnrc Monte Carlo simulations of the dose distribution from a 1 mm diameter narrow beam in a model of the film placed at six depths from 1.5 to 40 cm in polystyrene and solid water phantoms. Kernels for different area phantoms (50 x 50 cm2 and 25 x 25 cm2 polystyrene and 30 x 30 cm2 solid water) are produced. The Monte Carlo calculated kernel is experimentally verified with film, ion chamber and thermoluminescent dosimetry (TLD) measurements in polystyrene irradiated by a narrow beam. The kernel is then used in convolution calculations to, predict the film response in open and IMRT fields. A 6 MV photon beam and Kodak XV2 film in a polystyrene phantom are selected to test the method as they are often used in practice and can result in large energy-dependent artifacts. The difference in dose distributions calculated with the film kernel and the water kernel is subtracted from film measurements to obtain a practically film artifact free IMRT dose distribution for the Kodak XV2 film. For the points with dose exceeding 5 cGy (11% of the peak dose) in a large modulated field and a film measurement inside a large polystyrene phantom at depth of 10 cm, the correction reduces the fraction of pixels for which the film dose deviates from dose to water by more than 5% of the mean film dose from 44% to 6%.