Quality assurance tests for the Gamma Knife® Icon™ image guidance system

Abstract Introduction The Gamma Knife® Icon™ comes with an image guidance system for tracking patient motion and correcting for inter‐ and intrafractional shifts, mainly used with frameless thermoplastic immobilization. The system consists of a cone‐beam CT (CBCT) and a couch‐mounted infrared camera (IFMM). We report our quality assurance program for Icon's image guidance system. Methods The manufacturer‐provided tool is used for daily checks of CBCT positional precision. Catphan® phantom is used for monthly checks of CBCT image qualities (uniformity, contrast to noise ratio (CNR), and spatial resolution) for the two acquisition presets (low‐dose and high‐quality presets). On a semi‐annual schedule, we use a frame tool to check the agreement of CBCT‐based and Frame‐based stereotactic space coordinates by comparing the locations of five attached ball bearings in CT‐sim scans (Frame‐based coordinates determination) and in Icon's CBCT scans. On an annual basis, the accuracy of IFMM, image registration, and delivery‐after‐shift are tested using a translational stage. A weighted CT dose index is measured annually with a pencil chamber in CTDI head phantom. Results The CBCT precision check: 0.12 ± 0.04 mm (maximum deviations average). CBCT image quality: spatial resolution range: [6,7] lp/cm (low), and [7,8] lp/cm (high); uniformity: 12.82 ± 0.69% (low), and 13.01 ± 0.69% (high); CNR: 1.07 ± 0.08 (low), and 1.69 ± 0.10 (high). Agreement of CBCT‐based with Frame‐based stereotactic coordinates range: [0.33, 0.66] mm. Accuracy of IFMM: 0.00 ± 0.12 mm (average) with 0.27 mm (max.); image registration: 0.03 ± 0.06 mm (average) with 0.23 mm (max.); and delivery‐after‐shift: 0.24 ± 0.09 mm (average) with 0.42 mm (max.). CTDI w: 2.3 mGy (low), and 5.7 mGy (high). Conclusions The manufacturer‐required QA checks together with additional user‐defined checks are an important combination for a robust quality assurance program ensuring the safe use of Gamma Knife® Icon™'s image guidance and motion management features.


| INTRODUCTION
The high dose of radiation delivered to intracranial lesions in stereotactic radiosurgery (SRS) mandates precise localization. With current available technology, it is possible to localize targets with a noninvasive frameless system with the help of image guidance and motion management tools, as opposed to the traditional invasive frame system fixed to the skull. The Leksell Gamma Knife ® Icon™ (Elekta A.B., Stockholm, Sweden) comes equipped with an image guidance system that makes it feasible to use a frameless thermoplastic fixation mask instead of the traditional SRS invasive frame. The Icon comes with a Cone-beam CT (CBCT) system that can be used for defining the Leksell stereotactic space using imaging without the need for the traditional frame system. CBCT is also used in pretreatment for determining any translational and rotational shifts of the patient skull with respect to the reference CBCT image. For tracking intrafractional motions, the Icon comes with an Intra-Fraction Motion Management (IFMM) system consisting of an infrared (IR) camera that tracks the movement of a reflective marker, typically placed on the patient's nose, with respect to reference reflective markers permanently attached to the head rest adapter during treatment delivery. The CBCT and IFMM system specifications were described in previous publications. 1,2 A typical frameless patient treatment workflow starts with planning on nonstereotactic CT or MRI images that provide the needed anatomical information for treatment planning. A reference CBCT on the Icon system is taken with the patient head laying on a custom pillow and immobilized with the frameless thermoplastic system. This reference CBCT image is used to define the Leksell stereotactic coordinates and is coregistered with the planning CT/MRI image in Leksell GammaPlan (LGP) software (V11.0.2, Elekta Instruments, A.B., Stockholm, Sweden).
Prior to treatment delivery, a setup CBCT is taken in the treatment position and is co-registered with the reference CBCT to determine the interfractional shifts. Any translation or rotation differences between the setup CBCT and reference CBCT are calculated by the LGP software and an optimum translational couch shift is proposed. A new dose distribution plan is shown for physician approval taking into account the change in plan due to rotational/translational shifts. During the treatment delivery, the intrafractional motion is determined by tracking a reflector marker placed on the patient nose detected by the couchmounted IR camera. A displacement of this reflector above a threshold, that can be set from 0.5 to 3 mm, triggers an automatic delivery stop (in active mode). Intrafractional setup CBCTs can be acquired and coregistered with the reference CBCT if the marker displacement did not return back below the threshold, and the process of coregistration with the reference CBCT and plan adaptive is repeated. 3 In this study, we share our QA program for testing the image guidance features of the Gamma Knife Icon.

2.A | Manufacturer's required CBCT tests
The manufacturer requires two tests to be performed routinely by the user for the Icon's CBCT system: (a) CBCT Precision Test and (b) CBCT Image Quality Test.

2.A.1 | CBCT precision test
This is a console-programmed test to check the CBCT positional precision. The user scans the manufacturer-provided tool (QA Tool Plus) and the test algorithm detects the coordinates of four fiducial markers on the tool. The algorithm compares the coordinates with baseline values that were determined by a manufacturer's service calibration tool. The test algorithm also calculates the CBCT image volume and compares points with baselines to report the maximum deviation value. 4 This test is required by the manufacturer to be performed once a month but it is recommended to be carried out daily.
We report our daily maximum deviation values for a 2-yr period in this work.

2.A.2 | CBCT image quality tests
The second test required by the manufacturer is to check image qualities for the two scanning presets available on the Icon: the lowdose preset of nominal CTDI = 2.5 mGy and the high-quality preset of nominal CTDI = 6.3 mGy. The CBCT image quality test is required by the manufacturer to be performed on a semi-annual schedule, with the recommendations to be performed monthly. The Catphan ® 503 phantom (The Phantom Laboratory, Salem, NY) is CBCT scanned for this test and different modules within the phantom are utilized to check for spatial resolution; contrast to noise ratio (CNR); and image uniformity.
The spatial resolution was determined by an observer finding the highest numbered line pair that can be seen in a CBCT image of the phantom.
The CNR was calculated from an image of a polystyrene and LDPE inserts using the equation: where I PS and I LDPE indicate the mean pixel values using a 5-mm square image probe for the polystyrene and LDPE, respectively; and σ PS and σ LDPE indicate the standard deviation values for the same inserts.
For the uniformity test, five readings were taken at the center, 12, 3, 6, and 9 o'clock positions of an image of a homogeneous section of the Catphan 503 phantom using a 10-mm square image probe. The uniformity was calculated using the formula: where Ī max is maximum mean pixel value, and Ī min is the minimum mean pixel value of the five readings. 5 We report our 2-yr results of the manufacturer monthly image quality tests in this work.

2.B | User-defined image guidance tests
In our clinic, we developed additional independent tests to check the agreement of CBCT-based stereotactic space with Frame-based stereotactic space, the accuracies of the IFMM system, registration algorithm, and delivery-after-shift, and to monitor the CBCT dose consistency through weighted CT dose index (CTDI) measurements.
2.B.2 | Accuracy tests of IFMM system, registration algorithm, and delivery-after-shift For performing the accuracy tests, a treatment plan was created based on CT simulator images with two shots placed on the film planes just under the visible holes using the smallest collimator (4 mm). A reference CBCT was then taken with the Icon system and coregistered with the planning CT. A shift was introduced with the stage prior to the setup CBCT to mimic an interfractional motion, and the IFMM system was checked for reading accuracy against this shift.
A setup CBCT was taken next, and was coregistered with reference CBCT to calculate the shifts with the LGP registration algorithm which was checked against the actual introduced shifts. Two shift scenarios were tested in our work: (a) A small shift scenario (2 mm in x, 2 mm in y, and 2 mm in z directions); (b) A large shift scenario (20 mm in z, 5 mm in x, and 5 mm in y followed by a rotation of 1 degree).
The shots delivery was carried out for the same scenarios R is the rotation matrix, h z being the angle of rotation about the z-axis; R y ðh y Þ ¼ cosðh y Þ 0 sinðh y Þ 0 1 0 Àsinðh y Þ0 0 cosðh y Þ We report the results obtained during commissioning and two annual CTDI w measurements for both CBCT presets.

3.A | CBCT precision QA test results
The plot in Fig. 3    CBCT-based coordinateswith a maximum up to 0.8 mm. 9 We plan to perform this CBCT-based vs. Frame-based stereotactic space coordinates test on a semi-annual basis, with a tolerance limit of 1 mm.

3.B | CBCT image quality results
3.D | Accuracy tests results of IFMM system, registration algorithm, and delivery-after-shift Table 3  The registration algorithm test also showed accuracy in the order of 0.1 mm, with a maximum of 0.23 mm difference in the z direction. The agreement in the rotational shift was within 0.01 ± 0.01 degrees with a maximum of 0.03 degree.
We found the transformation matrix method we used to test the accuracies of the registration algorithm and the IFMM system to be a useful test with submillimeter accuracy. This method can easily be applied in clinical settings to check the registration algorithm, T A B L E 1 Summary results of monthly image quality tests (spatial resolutions, CNR, and uniformity) with the manufacturer specification for each test.

Test
[Range] or average ± SD Manufacturer specification CTDI 2.5 mGy (low-dose preset) CTDI 6.3 mGy (high-quality preset) Spatial resolution [6,7] lp/cm [7,8]  In clinical practice, one ambiguity the user faces is that the system-proposed translational corrections are not visible to the user in terms of updated couch positions after the co-registration of setup CBCT with the reference CBCT to account for patient interfractional shifts, and thus the user is left to trust the system automatic application of these shifts without any verification. We feel that testing the unit's determination of shift and applying it correctly is an important routine QA needed to ensure the safe use of the system. We plan to perform these accuracy tests of IFMM, registration, and deliverafter-shift on annual bases. Taking into consideration the limitations of our testing tool and methodology, we set a 0.5 mm tolerance criterion for each test, and a 0.5 degree for the registration algorithm test. T A B L E 3 Summary results of accuracy tests of the Intra-fraction Motion Management system (IFMM), registration algorithm, and deliveryafter-shift to a film. The results for IFMM and registration (3D stage method) and the shot measurements are averaged over small (2 mm in each of x, y, and z direction) and large (5 mm in x and y directions, 20 mm in z direction, and 1 degree rotation along y) shift test scenarios. The difference is calculated as measured value minus expected value. unit are expected, and the differences are dependent on the generator, tube output, and manufacturing of the unit covers. 5 We have established our tolerance criteria for this test to be ±5% of our CTDI w baselines, and plan to continue performing this test on an annual basis as recommended by AAPM Task Groups 142 7 and 179. 8

| CONCLUSION
In this study, we combined the manufacturer-required routine QA checks with additional user-defined checks for a comprehensive and robust quality assurance program for the Gamma Knife Icon's image guidance system ensuring its safe use. We found our system performance to meet the manufacturer specifications and our set limits, and to be comparable to other reported values in the literature.

CONF LICT OF I NTEREST
The authors have no conflicts of interest to disclose related to the topic of this work.