Dosimetry of a novel rotating gamma system for stereotactic radiosurgery

Abstract

This is the first report of the basic dosimetric properties of a new rotating gamma system: the RGS Vertex360^™. Dosimetric properties were compared to those measured with traditional rotating gamma systems and with the Leksell Gamma Knife. The RGS Vertex360 is similar to the original rotating gamma system developed by OUR New Medical Technology Development Co., Ltd. (Shenzen, China), however, there are a few notable differences including the angular arrangement of the sources. Basic dosimetric properties of the RGS Vertex360 were measured including: absorbed dose rate, output factors, mechanical and radiation center accuracy and dose profiles. A significant discrepancy was observed for the 4 mm output measured from the RGS Vertex360 compared to those obtained from previous rotating gamma units: the 4 mm output from the RGS Vertex360 (0.807) was 32-38% higher than those measured from previous units. This is somewhat surprising considering the excellent agreement in 4 mm outputs from the RGS Vertex360, the corresponding outputs specified by the manufacturer of the original OUR unit and those measured for the Leksell Gamma Knife. The mechanical accuracy was similar to previous rotating gamma systems while the 50-90% penumbra was narrower. Dose profiles compared favorably with the Leksell Gamma Knife: in many instances the measured penumbra was narrower for the RGS Vertex360.

Notwithstanding the 4 mm output factor, the dosimetric properties of the RGS Vertex360 compared favorably with those of previous rotating gamma systems. The 4 mm output discrepancy was attributed to suboptimal alignment of the primary and secondary collimators in previous studies. The dosimetric properties of the RGS Vertex360 and the Leksell Gamma Knife were similar and, taken together, the results suggest that the new rotating gamma system is well suited for stereotactic radiosurgery procedures.

INTRODUCTION

The gamma knife was developed in 1967 by Lars Leksell and Borje Larsson for the treatment of movement disorders and intractable pain. [1,2] By the mid 1970s, Swedish neurosurgeons recognized the potential of the device for the treatment of intracranial tumors and vascular malformations. [3] The Leksell Gamma Knife (LGK) is manufactured and marketed by Elekta Instruments, Inc. (Stockholm, Sweden). The three current LGK models deliver focused radiation via 201 (Models 4 and 4C) or 192 (Perfexion^®) highly collimated ⁶⁰Co sources. The traditional units (4 and 4C) consist of fixed radiation sources, a moveable couch and four collimator helmets defining beam diameters of 4, 8, 14 and 18 mm. Since the collimator is attached to the couch, the patient must be released from the positioning system during collimator changes. This contributes to an overall increase in treatment time. In the Perfexion^® system, 3 collimators rotate within the unit and can be changed automatically via computer control resulting in a significant diminution in overall treatment time. Other advantages of the Perfexion^® system include expanded cranial reach, automatic beam blocking, dynamic shaping and advanced dose planning software facilitating composite shots. [3]

Rotating gamma systems (RGSs) have recently emerged as an alternative to traditional gamma knife units. The first RGS (OUR New Medical Technology Development Co., Ltd., Shenzen, China) was installed in China in 1996 and soon thereafter (May, 1997), the RGS received U.S. FDA approval. [4] The RGS consists of 30 ⁶⁰Co sources distributed symmetrically around a hemispherical shell capable of rotating at a speed of 1 to 4 rpm. [5] Within the source hemisphere, a secondary collimating hemisphere rotates with the sources. The collimating hemisphere contains six groups of five collimator holes arranged in an identical fashion as the sources. Through selection of a particular group of collimator holes that can be aligned with the sources, different beam diameters can be chosen. [3] As with the Perfexion^® system, collimators are changed automatically, however, RGS treatments are accomplished with only 30 sources resulting in a significant overall cost savings. In addition, the RGS can simulate an infinite number of beams and is capable of intensity-modulated radiation surgery (IMRS.) In 2000, an American-based company (American Radiosurgery, Inc., San Diego, CA) began manufacturing, marketing and selling RGSs. The system was originally called the RGS GammaART-6000, but in 2008, it was renamed the RGS Vertex360 to more accurately reflect its 360 degree rotational capabilities. There are currently four American Radiosurgery systems worldwide, two of which are in the U.S.

An RGS Vertex360 unit was recently installed at the Rotating Gamma Institute Orange County (Anaheim, CA). The dosimetric and mechanical properties of this system (RGS_OC) was evaluated and compared to two other units: the original Chinese OUR RGS (RGS_C) installed at the Auhai Radiosurgery Center in Beijing, China, [4] and a unit installed at the University of California, Davis Cancer Center (RGS_U). [6] Particular attention was devoted to the output measurements since previous studies found that the 4 mm collimator outputs from the RGS_C and RGS_U significantly underestimated both the manufacturer’s specification and the LGK outputs.

MATERIALS AND METHODS

The Rotating Gamma System

The RGS_OC (Figures 1 and 2) consists of the radiation unit with shielding housing and protective doors, rotating source carrier, rotating collimator system, patient support system and the electronic control system. The total initial activity of the 30 ⁶⁰Co sources is 2.22 x 10¹⁴Bq (± 10%). The unit is similar to the OUR Model RGS system described in Goetsch et al. (1999), however, there are a few notable differences including the source arrangement. In the OUR system, the sources are arranged symmetrically in 30 concentric circles and grouped in 6 spirals spaced at 60^o intervals about the axis of rotation. In contrast, the 30 sources in the RGS_OC system are located in one segment of the source carrier (Figure 3). As illustrated in Figure 4, the sources in the RGS_OC unit subtend a solid angle from 13 to 53^o with respect to the vertical (in intervals of 1^o) compared to 14 to 43^o in the OUR system. Additionally, the RGS_OCutilizes IMRS which allows the sources to be periodically turned on and off, thus excluding pre-selected directions from which radiation can enter the skull. This has an effect similar to plugging in a static source device. The addition of IMRS therefore allows plugging in arbitrary directions yielding improved dose distributions.

Figure 1.

American Radiosurgery RGS Vertex360 installed at the Rotating Gamma Institute Orange County.

Figure 2.

Schematic of the source housing. The numbers correspond to locations where leakage radiation measurements were made.

Figure 3.

Source arrangements in the RGS_OC and OUR RGS. The secondary collimator depicts the hole arrangements for the RGS_OC.

Figure 4.

Angular source arrangement associated with the RGS_OC. Sources are arranged within the primary collimator hemisphere at latitude angles between 13 and 53^o as measured from the sagittal plane spanning the opening of the hemisphere. The gamma angle (r) is adjustable to allow for aligning of the patient’s longitudinal axis in relation to the system’s rotating axis.

The relationship between the primary and secondary collimators is illustrated in Figure 5. The rotating source carrier is a hemispheric shell containing the 30 cobalt sources. The source carrier also serves as the primary collimator focusing the rays to the center of the hemisphere. A second coaxial hemispheric shell, the secondary collimator body, is located inside the source/ primary collimator. The secondary collimator contains five groups of collimator holes arranged in the same helical fashion as the sources: one group of holes is blocked with shielding plugs while the remaining four groups consist of holes of varying diameters (4, 8, 14 and 18 mm). Different focal spot sizes can be achieved simply by selecting which group the primary collimators are aligned with. When the system is not in use, all sources are aligned with the “dead block” positions on the secondary collimator body. At the start of treatment, the sources will align with the designated group of collimator holes through differential rotation of the two sets of collimators. When the appropriate hole-source alignment has been achieved, the primary and secondary collimators rotate synchronously during treatment resulting in 30 non-overlapping 360^o arcs.

Figure 5.

Relationship between the primary and secondary collimators.

The treatment planning system (Explorer 4D^TM; American Radiosurgery, Inc., San Diego, CA) is novel and specific to the RGS Vertex360. The system is compatible with a UNIX^TM-based Mac OS X operating system and runs on the OsiriX^TM medical image viewing platform. The system is capable of 3D and 4D time-domain reconstruction and stereoscopic viewing as well as imaging (CT, MRI and PET) fusion and automatic skin and skull shape detection.

Determination of Absorbed Dose Rate

Measurement of absorbed dose rate was accomplished with a Capintec Model PR-05P “mini” chamber connected to a Capintec Model 192 Exposure Rate Meter. Both chamber and meter were calibrated at the MD Anderson Calibration Laboratory. The chamber was inserted into the center of a 16 cm diameter spherical polystyrene phantom which was placed in the stereotactic frame positioning system. The center of the phantom was positioned at the intersection point of the beams: (x, y, z) = (100, 100, 100). The phantom was exposed using the 18 mm collimator and the absolute dose rate was determined using the TG-21 formalism.

Output Factors and Dose Delivery Accuracy

Collimator output factors were measured with the ionization chamber and rate meter described previously. As a verification check, measurements were also taken with Gafchromic MD-55 film. This is particularly important for the smallest collimators (4 and 8 mm) since the small beams are unlikely to intercept the entire ionization chamber volume thus resulting in low readings. The film was sandwiched between two 5-mm thick flat plastic plates which fit into a slot made in the spherical polystyrene phantom. Two films were exposed for each of the four collimators: one on the XY plane and the other on the XZ plane. Each film was exposed to 15 Gy and the film density was determined at the University of Wisconsin (UW) ADCL. In each case (film and chamber), outputs were normalized to those obtained with the 18 mm collimator.

Dose delivery accuracy was checked on a monthly basis by measuring the output from the 18 mm collimator and comparing the result to the original output (corrected for ⁶⁰Co decay) determined at the time of commissioning. Dose accuracy for the 4, 8 and 14 mm collimators was determined monthly by measuring the outputs and calculating the output factor (the ratio of the output of each collimator to that measured from the 18 mm collimator). The three output factors were then compared to the original ratios determined at the time of commissioning. Linearity checks were also performed monthly from measurements of exposure as a function of time (1, 2, 5 and 10 min.) for each collimator.

Mechanical and Radiation Center Accuracy

Coincidence measurement of mechanical and radiation isocenters was accomplished using a vendor supplied aluminum pinprick device. The device orients the film in the plane perpendicular to the axis of rotation (XY plane) and allows a pinprick to be placed at the mechanical isocenter of the device. The exposed film profiles show a sharp dip where the needle has penetrated the film. This allows the calibration laboratory to estimate the distance from the center of the radiation field to the mechanical isocenter. [4] Two films were exposed using the 4 mm collimator: one in the XY plane and the other in the XZ plane. Films were read at the UW ADCL.

Two methods were used to determine the coincidence error for the mechanical and radiation field centers. In the best-fit ellipse method, a profile is constructed by analyzing the optical density along a slice through the center of the scanned image on the film and graphing the optical density as a function of position. In this method, an ellipse is fitted to the scanned image at the half maximum optical density value and the center of the ellipse is superimposed on the image with open cross hairs. The inner edge of each cross hair is 0.4 mm from the center of the ellipse which corresponds to the recommended maximum offset as specified by the AAPM. The distance to the center of the pin prick can be directly measured in each dimension and the values added in quadrature to obtain the total position error of the mechanical center and the radiation field isocenter.

The orthogonal profile method determines the center of a dose profile for each dimension and compares the position of the center of the profile to the position of the center of the pin prick placed on the film prior to exposure. The three orthogonal distances obtained with this method were then added in quadrature to determine the total position error of the mechanical center and the radiation field isocenter.

Dose Profiles

A total of eight Gafchromic MD-55 films were exposed to determine radiation field profiles. Two films were exposed for each of the four collimators to determine profiles in both XY and XZ planes. The placement of the films in the spherical phantom was identical to that previously described. Each film was exposed to approximately 15 Gy. In all cases, films were analyzed at the UW ADCL. Dose profile and mean radial profile plots were constructed using the best-fit ellipse and orthogonal profile methods. In addition, dose contour plots were generated from the image files. Computer generated images for each film were produced with isodose contour lines (normalized to 100% image density) superimposed over the image.

Timer Accuracy

Timer accuracy was measured against a stopwatch. During the 100 s exposure, the stopwatch started when the treatment console’s digital clock indicated time at t. The watch was stopped when the digital clock indicated t + 100 s.

Leakage Radiation Survey and Safety Interlocks

Leakage radiation was measured around the RGS_OC with a Fluke 451 ion chamber survey meter. The survey was performed in two parts: outside the treatment room (5 ft. above the floor and a few inches from the closed treatment door) with the unit in treatment mode using 18 mm collimators, and inside the room at distances of approximately 1 m from the location of the source cluster with the source shield door closed and the sources aligned with the “dead block” positions. The door interlock, emergency off button, audiovisual system and radiation warning light were tested.

RESULTS

Absolute Dose Rate

Ionization chamber measurements, followed by application of the TG-21 protocol, yielded a dose rate output of 2.93 Gy min^-1 in the center of the 16-cm-diameter polystyrene phantom using the 18 mm collimator. This is well within the acceptable tolerance of 3 ± 0.3 Gy min^-1. The measured dose rate was subsequently entered into the treatment planning system.

Output Factors and Dose Delivery Accuracy

Collimator output factors are summarized in Table 1. For comparative purposes, output factors from the RGS_C [4] and RGS_U [6] are presented. Results of output measurements from the RGS_OC showed good agreement between ionization chamber and film for the two largest collimators, however, the chamber significantly underestimated output factors for the two smallest collimators. This is particularly evident in the case of the 4 mm collimator, where the measured output was 40% lower than that determined from film measurements. The origin of this discrepancy was attributed to the large chamber volume-to-beam diameter ratio.

Table 1.

Collimator output factors for three RGSs.

	RGSC		RGSU		RGSOC
Collimator (mm)	PTW chamber*	A14SL chamber†	TLD†	Gafchromic film†	Capintec chamber	Gafchromic film
18	1.000	1.000	1.000	1.000	1.000	1.000
14	0.982	0.975	0.984	0.980	0.978	0.973
8	0.937	0.861	0.912	0.844	0.878	0.908
4	0.610	0.604	0.593	0.583	0.484	0.807

^*Reference #4

^†Reference #6

Due to the uncertainties associated with the ion chamber measurements, output factors determined from film measurements were used in all patient treatments. The monthly ion chamber output measurements thus served as a constancy check. The dose accuracy of the 18 mm collimator was found to be within 0.5% while the accuracy of the 4, 8 and 14 mm collimators were within 1% based on comparisons with the outputs determined at the time of machine commissioning. The dose was found to be linear to within 1% for all four collimators. As shown in Table 1, there was a significant discrepancy between the 4 mm output factor measured with the RGS_OC (0.807) and those measured with the two other RGSs (0.583 – 0.610). The 4 mm output factor measured for the RGS_OC is similar to the manufacturer’s specification for the RGS_C (0.881; 4) and is in close agreement with typical LGK outputs which range from 0.82 – 0.9. [7]

Mechanical Accuracy

Results of the pinprick films are summarized in Table 2. The average mechanical accuracy of the RGS_OC as determined by the best fit ellipse and orthogonal profile methods was approximately 0.2 mm. As shown in Table 2, this is in good agreement with the accuracy of the RGS_C and RGS_U and is well within the manufacturer’s specified tolerance of 0.5 mm. Dose profiles in the XY and XZ planes are illustrated in Figure 6.

Table 2.

Mechanical accuracy of three RGSs

RGSC*	RGSU†	RGSOC
0.23 mm	X and Y: 0.29 mm	0.18 mm (best fit ellipse)
(4 and 8 mm)	Z: 0.75 mm	0.21 mm (orthogonal profile)
	(4 mm only)	(4 mm only)

^*Reference #4

^†Reference #6

Figure 6.

Mechanical accuracy of the RGS_OC. Pinprick results obtained with the 4 mm collimator in the XY (a) and XZ (b) planes.

Dose Profiles

Film isodose plots from the UW-ADCL are shown in Figures 7a and b, and 8a and b. Representative results are shown for the smallest (4 mm) and largest (18 mm) collimators in the XY and XZ planes. The isodose curves in Figure 7a are circular while those in Figure 7b demonstrate horn-like extensions. These horns are created by the beam entering and leaving the horizontal film at angles between 13 and 53^o from the vertical axis. Due to the larger beam, the horns are not as prominent for the 18 mm collimator (Figure 8b). No horns are observed for the vertically placed films since the beam contribution to the film is symmetric about left-right or anterior-posterior. Beam profile scans are shown in Figures 7c and d, and 8c and d. In the XY plane, the X and Y profiles are almost identical for both collimators (Figs. 7c and d). In contrast, the Z axis profiles are narrower than the X and Y profiles (Figs. 8c and d).

Figure 7.

Isodose contours with the 4 mm collimator film exposure on the XY (a) and XZ (b) planes. Distances denote the full width at 50% isodose profile in the specific directions. Isodose lines from 90 to 10 % (in 10% increments) are shown. Corresponding collimator profiles on the X and Y axes and the X and Z axes are illustrated in (c) and (d), respectively.

Figure 8.

Isodose contours with the 18 mm collimator film exposure in the XY (a) and XZ (b) planes. Corresponding collimator profiles on the X and Y axes and the X and Z axes are illustrated in (c) and (d), respectively.

Profile diameters at the 90 and 50% isodose lines, and the resultant penumbra (the difference in the full widths between the two isodose lines) are summarized for all four collimators in Table 3. For comparative purposes, data are also provided for the RGS_C, and the Gamma Knife (Model C; Good Samaritan Hospital, Los Angeles, CA). Results show that the 90-50 penumbra (Pen.) obtained with the RGS_OC are smaller than those measured from the RGS_C for all collimators and in both planes. RGS_OC penumbra compare favorably with those measured from the LGK, in fact, RGS_OC penumbra are smaller (or equivalent) to LGK penumbra for the 14 and 18 mm collimators.

Table 3.

Measured field diameters for two RGSs and a LGK.

		RGSC*			LGK†			RGSOC
Coll./plane	90%	50%	Pen.	90%	50%	Pen.	90%	50%	Pen.	Diff1	Diff2
4 XY	3.0	6.7	3.7	3.8	7.2	3.4	3.8	7.3	3.5	-0.2	0.1
4 XZ	2.7	6.3	3.6	3.5	6.9	3.4	3.5	6.9	3.4	-0.2	0.0
8 XY	6.3	12.0	5.7	7.8	12.8	5.0	7.8	12.7	4.9	-0.8	-0.1
8 XZ	6.2	11.4	5.2	7.3	12.0	4.7	7.0	11.9	4.9	-0.3	0.2
14 XY	11.7	18.5	6.8	13.7	19.6	5.9	13.5	19.4	5.9	-0.9	0.0
14 XZ	11.1	17.4	6.3	13.5	19.3	5.8	13.4	18.9	5.5	-0.8	-0.3
18 XY	15.7	23.1	7.4	17.2	24.4	7.2	17.0	23.9	6.9	-0.5	-0.3
18 XZ	14.9	22.1	7.2	16.9	23.8	6.9	17.0	23.3	6.3	-0.3	-0.6

^*Reference #4

^†Measurement made at Good Samaritan Hospital, Los Angeles, CA.

Diff1 is the difference in penumbra between the RGS_OC and the RGS_C

Diff2 is the difference in penumbra between the RGS_OC and the LGK

TIMER ACCURACY

The difference between the treatment console digital clock and the stopwatch over a 100 s count was 0.2 s. Therefore the timer accuracy was of the order of 0.2%.

Leakage Radiation and Safety Interlocks

A total of 14 leakage measurements were made at distances of approximately 1 m from the treatment unit. All measurements were made with the shielding door closed and all sources in the “dead block” position. As shown in Table 4, leakage measurements ranged from 0.02 to 1.59 mR h^-1. These values are slightly lower than those obtained by Kubo et al. [6] at similar locations around the RGS_U. Leakage measurements were also performed outside the treatment room with the unit in treatment mode. The highest reading (0.12 mR h^-1) was recorded at the closed treatment door. All other measurements (operating console, doctors’ offices, patient waiting room) were identical to background (approximately 0.01 mR h^-1).

Table 4.

Results of leakage radiation measurements (mR h^-1). Location numbers are defined in Figure 2.

Location	1	2	3	4	5	6	7	8	9	10	11	12	13	14
	0.02	N/A	0.09	0.91	0.05	0.03	0.03	0.03	0.02	0.03	0.03	1.59	N/A	0.03

The door interlock, emergency off button, audiovisual system and radiation warning light were found to be in working order.

DISCUSSION

Accurate dosimetry of highly conformal radiation delivery devices, such as the RGS_OC, is of particular importance in the treatment of brain lesions which require high doses and sub mm accuracy. In that context, the most relevant finding of this study was the high output factor measured from the 4 mm collimator. The measured value was in good agreement with both the manufacturer’s specifications for the original unit (RGS_C) and with typical LGK outputs. In comparison, the 4 mm output factors measured from two previous RGSs (Table 1) were 32-38% lower than the RGS_OC value determined with Gafchromic film. The reason for this discrepancy is unclear: it may be due to suboptimal alignment of the primary and secondary collimator hemispheres. In the case of the smallest collimator (4 mm), even a slight misalignment can result in a considerable decrease in the output factor. Such misalignment would also impact the dose profiles and may therefore explain the larger penumbra observed with the RGS_C (Table 3.)

From a clinical standpoint, higher output factors are desirable as overall treatment time is minimized. This is particularly important for the treatment of functional disorders such as trigeminal neuralgia and tremors which are commonly treated with 4 mm collimators and require single fraction doses of up to 80 and 150 Gy, respectively. In these cases, optimized collimator alignment also produces a high degree of dose conformality due to the reduced penumbra.

In order to optimize the collimator alignment, the unit was placed in “service” mode which allows the operator to control the position of the secondary collimator hemisphere with respect to the stationary primary source hemisphere. For a given collimator size, the primary hemisphere was held stationary while the secondary hemisphere was rotated in 0.5^o increments. For each angle, the output was measured with the ionization chamber and rate meter as previously described. This procedure was repeated until the maximum output was recorded. The position of the secondary collimator hemisphere with respect to the primary source hemisphere was entered into the system and this was considered to be the optimum alignment for that particular collimator. This optimization procedure was repeated for each of the four collimators.

CONCLUSIONS

Overall, the RGS_OC compares favorably with previous RGSs: the mechanical accuracy is similar while the dose profiles are superior as illustrated by the narrower 50-90% penumbra. Perhaps most importantly, the 4 mm collimator output factor has been optimized resulting in significant reduction in treatment times. The output factors of the RGS_OC are similar to those of the LGK and, in some cases, the dose profiles are better as evidenced by narrower penumbra. Taken together, the results provide a high degree of confidence for the clinical suitability of the RGS_OC.