Assessment of the accuracy and stability of frameless gamma knife radiosurgery

Abstract The aim of this study was to assess the accuracy and stability of frameless gamma knife radiosurgery (GKRS). The accuracies of the radiation isocenter and patient couch movement were evaluated by film dosimetry with a half‐year cycle. Radiation isocenter assessment with a diode detector and cone‐beam computed tomography (CBCT) image accuracy tests were performed daily with a vendor‐provided tool for one and a half years after installation. CBCT image quality was examined twice a month with a phantom. The accuracy of image coregistration using CBCT images was studied using magnetic resonance (MR) and computed tomography (CT) images of another phantom. The overall positional accuracy was measured in whole procedure tests using film dosimetry with an anthropomorphic phantom. The positional errors of the radiation isocenter at the center and at an extreme position were both less than 0.1 mm. The three‐dimensional deviation of the CBCT coordinate system was stable for one and a half years (mean 0.04 ± 0.02 mm). Image coregistration revealed a difference of 0.2 ± 0.1 mm between CT and CBCT images and a deviation of 0.4 ± 0.2 mm between MR and CBCT images. The whole procedure test of the positional accuracy of the mask‐based irradiation revealed an accuracy of 0.5 ± 0.6 mm. The radiation isocenter accuracy, patient couch movement accuracy, and Gamma Knife Icon CBCT accuracy were all approximately 0.1 mm and were stable for one and a half years. The coordinate system assigned to MR images through coregistration was more accurate than the system defined by fiducial markers. Possible patient motion during irradiation should be considered when evaluating the overall accuracy of frameless GKRS.

month with a phantom. The accuracy of image coregistration using CBCT images was studied using magnetic resonance (MR) and computed tomography (CT) images of another phantom. The overall positional accuracy was measured in whole procedure tests using film dosimetry with an anthropomorphic phantom. The positional errors of the radiation isocenter at the center and at an extreme position were both less than 0.1 mm. The three-dimensional deviation of the CBCT coordinate system was stable for one and a half years (mean 0.04 AE 0.02 mm). Image coregistration revealed a difference of 0.2 AE 0.1 mm between CT and CBCT images and a deviation of 0.4 AE 0.2 mm between MR and CBCT images. The whole procedure test of the positional accuracy of the mask-based irradiation revealed an accuracy of 0.5 AE 0.6 mm. The radiation isocenter accuracy, patient couch movement accuracy, and Gamma Knife Icon CBCT accuracy were all approximately 0.1 mm and were stable for one and a half years. The coordinate system assigned to MR images through coregistration was more accurate than the system defined by fiducial markers. Possible patient motion during irradiation should be considered when evaluating the overall accuracy of frameless GKRS.

| INTRODUCTION
Frameless gamma knife radiosurgery (GKRS) can be performed with the latest gamma knife model, Gamma Knife (GK) Icon TM [ Fig. 1(a)]. A stereotactic coordinate system necessary for frameless GKRS is assigned by coregistration of clinical magnetic resonance (MR) and/or computed tomography (CT) images with cone-beam computed tomography (CBCT) images obtained with a CBCT system. After a treatment plan is completed based on this coordinate system, the patient is fixed by a mask, and another set of CBCT images is obtained such that the deviations in patient position can be automatically corrected. During irradiation, the motion of the patient is monitored by a high-definition motion management (HDMM) system. When the motion is greater than a predefined limit for more than 2 s, the sources are retracted to their temporary storage positions. If the motion is maintained at levels greater than the limit for 30 seconds, the patient couch moves out, and CBCT imaging is repeated to correct for the movement. The overall accuracy of the frameless GKRS depends on numerous factors that can be classified into three groups: radiation-related factors, imagerelated factors, and patient motion. Given that the accuracy of framebased GKRS, such as the absolute dose measurement, treatment planning program accuracy, frame accuracy, and MR image distortion, has been studied in many previous works; 1-12 this work focused primarily on the accuracy aspects unique to frameless GKRS. The accuracy and stability of the radiation isocenter, the accuracy of patient couch movement, the accuracy and stability of CBCT images, the accuracy of image coregistration, and the accuracy of the HDMM system were assessed. whole procedure test of the geometrical accuracy of frameless irradiation with CT images were also performed. The stability of the measured values over one and a half years after installation was also analyzed. During the same period, independent periodic management was performed by the manufacturer at 6-month intervals. The assessment of patient motion during irradiation was excluded in this work because such motion is patient-specific and cannot be evaluated in phantom studies. The results were compared with the accuracy and image quality obtained at the time of machine commissioning 13 and with the short-term stability of the CBCT system. 14

2.A | Accuracy of the radiation isocenter and patient couch movement
The radiation isocenter of a gamma knife should coincide with the center of the patient-positioning system (PPS). The deviation between the radiation isocenter and the center of the PPS was measured using radiochromic films (GafChromic TM EBT3; Ashland Specialty Ingredients, NJ, USA) following a vendor-recommended procedure described by Novotny et al. 15 In brief, a film was set within a special tool provided by the manufacturer, and the center of the PPS was marked by a sharp pin. Then, the film was irradiated using the 4-mm collimator of a gamma knife and scanned. The deviation of the radiation isocenter, defined as the difference between the center of the radiation peak and the PPS center marked on the film, was measured. Three films were used in each test, and the procedure was performed biannually and results of four measurements were analyzed. In the stereotactic coordinate system of GKRS, the radiation isocenter corresponds to a point of x = y = z = 100.0, where the x-axis is defined from the right to the left of a patient, the y-axis is defined from the posterior to the anterior, and the z-axis is defined from the head to the feet. To evaluate the accuracy of patient couch movement, the same film test was repeated at an extreme position (40.0, 160.0, 100.0). The deviation of the radiation isocenter was also assessed daily using a diode detector installed on a vendor-provided tool [ Fig. 1(b)]. After measuring the dose distributions of the 4-mm collimator along each axis, the radiation isocenter was automatically determined by the control system as the middle point of two full width at half maximum points, and deviations from 100.0 were recorded with 0.1-mm precision.

2.B | Accuracy of CBCT images
The CBCT of the GK Icon obtains images by rotating the C-arm by 197°because the arm cannot pass through the patient couch [ Fig. 1(a)]. The source-to-detector distance was 1000 mm, and the source-to-axis distance was 790 mm. The cone beam angle was 15°, where I PS is the mean pixel value measured in a square of interest of size 5 9 5 mm 2 located in the polystyrene insert, I LDPE is the mean pixel value measured in the LDPE insert, and r PS and r LDPE are standard deviations in each region for the polystyrene and LDPE, respectively. The uniformity of the CBCT images was calculated using Eq. (2): where I high and I low are the highest and lowest mean pixel values measured in five regions of interest 10 9 10 mm 2 in size located at the center and at four points 45 mm from the center along the left, right, anterior, and posterior directions. The uniformity was measured for an image obtained at the homogeneous portion of the phantom. The image resolution, CNR, and uniformity were obtained twice a month.

2.C | Accuracy of image coregistration
The accuracy of stereotactic coordinates assigned by coregistration of the GK Icon CBCT images with clinical CT or MR images was evaluated using the commercial phantom CIRS 603a (CIRS Inc., Norfolk, VA, USA

2.D | Whole procedure test
To examine the overall geometrical accuracy of frameless GKRS, whole procedure tests were performed three times using EBT3 films and an anthropomorphic phantom, CIRS 605 (CIRS Inc., Norfolk, VA, USA). After a film was set in the phantom, clinical CT images were obtained. Then, the phantom was fixed with a mask as shown in

2.E | Accuracy of the HDMM system
In frameless GKRS, a marker is attached on the patient nose, and the motion is monitored by the HDMM system during irradiation.
The HDMM system consists of an infrared stereoscopic camera, four reference markers, and one patient marker. The infrared camera is mounted onto an arm on the patient couch and continuously tracks the movement of the patient marker during treatment. The patient is immobilized with a thermoplastic mask, and the motion of the patient marker is tracked at a frequency of 20 Hz, with an accuracy of 0.15 mm. 16 The accuracy of HDMM was verified using an inhouse device (Fig. 4). By moving a cube along the x-or z-axis using a depth micrometer (Mitutoyo Corp., Kanakawa, Japan), the HDMM values were recorded. Five measurements were executed in each direction. The squared HDMM values were fitted with a secondorder polynomial of the micrometer movements.

2.F | Statistical analysis
All statistical analyses were performed using the commercial package IBM â SPSS â Statistics version 22 (IBM Corp, Armonk, NY, USA).
Pearson correlation analysis was used to investigate the relationship between two variables, and an independent t test or one-way ANOVA was used to compare mean values. When the P value was less than or equal to 0.01, the difference was accepted as statisti-

3.B | Accuracy of CBCT images
Analysis of the one-dimensional deviations of the four ball bearings showed that the one-dimensional deviations along each axis were closely correlated (p < 0.001), indicating that the CBCT images acted as a rigid body. Figure 6  than that of their system (9.2% and 8.8%, respectively, for each preset). 13

3.C | Accuracy of image coregistration
When stereotactic coordinate systems were defined using the fiducial markers, the mean fiducial marker registration error was 0.2 mm for the CT images and 0.4 mm for the MR images. The maximum registration error was 0.5 mm and 1.0 mm for the CT and MR images, respectively. The coordinate system of the CBCT images was used as the reference system in this work because its geometrical deviations were less than 0.12 mm, as described in the previous section. The coordinates of each point consistently exhibited a standard deviation of 0.1 mm. Image coregistration could be repeated with differences of the same order as the measurement error, that is, with a standard deviation of 0.1 mm. with the CBCT images, their coordinates were more accurate than those assigned by the fiducial markers (p < 0.001). This finding suggests that frameless GKRS with coregistered MR images may achieve better imaging accuracy than frame-based GKRS. The errors in the fiducial marker-based MR coordinates were primarily due to image distortion, marker location error, and marker registration algorithms. 18 The MR image distortion, that is, the mean deviation of the | 153 larger errors than frameless GKRS with CT images because their image coregistration error was increased, the whole procedure error is still expected to be at the submillimeter scale because minimal differences were noted between the MR and CT coregistration error (0.4 mm vs 0.2 mm).

3.E | Accuracy of the HDMM system
The proportional coefficient for the micrometer reading and the movement provided by the HDMM system was 1.00 AE 0.03 for both the x-and z-axes, indicating that the HDMM system accurately measured the cube movement. Figure 8 presents the measurement data and their fitted lines. The isolated line in the x-axis movement was attributed to the fact that its reference point was different from that of the other measurements. Of note, the accuracy of HDMM does not guarantee the accuracy of irradiation. The operator at irradiation sets the limit of allowed patient motion, and the actual movement range differs from patient to patient. Thus, the final accuracy of irradiation is different for each patient.

| CONCLUSION
The positional accuracy of frameless GKRS was measured in the submillimeter range. The radiation isocenter accuracy, patient couch movement accuracy, and GK Icon CBCT accuracy were all approximately 0.1 mm and were stable for one and a half years. The coordinate system assigned to MR images through coregistration was more accurate than the system defined by fiducial markers. Possible patient motion during irradiation should be considered when evaluating the overall accuracy of frameless GKRS.