Validation of MLC-based linac radiosurgery for trigeminal neuralgia.

This study details a validation process for linear accelerator-based treatment of trigeminal neuralgia using HD-MLC field collimation. Nine trigeminal neuralgia treatment plans utilizing HD-MLC were selected for absolute dose measurement at isocenter using a commercial scintillating detector in an anthropomorphic phantom. Four plans were chosen for film dosimetry measurements in each of the three principal planes to assess spatial dose distribution agreement with the treatment planning system. Additionally, trajectory log analysis for each treatment field in the nine cases was performed to assess mechanical positioning accuracy of the MLC system during delivery. Scintillator and film measurements both revealed mean dose agreement at isocenter of better than 3% while FWHM of the 2D dose distribution in each plane showed agreement between plan and measurement within 0.2 mm. Analysis of log files revealed a maximum MLC leaf positioning error of 0.04 mm across 178 treatment fields. In conjunction with a quality-controlled treatment delivery methodology, an appropriately commissioned treatment planning system can be used for accurate and clinically appropriate design of trigeminal neuralgia treatment plans utilizing HD-MLC.


| INTRODUCTION
After more than five decades of development, stereotactic radiosurgery (SRS) is a widely available treatment technique for intracranial tumors using both dedicated units like Gamma Knife and CyberKnife, as well as linear accelerators (linacs). 1 SRS uses finely collimated beams to deliver ablative doses in a single or few fractions to treat benign and malignant tumors in the brain. SRS delivery on linacs became possible only after refinements in mechanical precision of the delivery system. Early investigators of linacs as a radiosurgery tool constructed special hardware that attached to the accelerator and allowed a higher degree of precision to be achieved than was possible with standard linac treatments. 2 The use of a tertiary collimating system with circular apertures closer to the patient on a linac reduces both the beam penumbra and the susceptibility to positioning error. 3 Modern linacs for SRS delivery from vendors come with both circular collimator attachments and a computer-controlled multi-leaf collimating (MLC) system which is integrated in the head of the machine to shape the beam. The user has the option of using either beam shaping device depending on the application. with finer leaf widths (2.5-3.0 mm) for the purpose of treating very small targets. 4,5 With the improved technology, modern machines can achieve overall couch/gantry/collimator isocentric accuracy within 0.6 mm. 6 When these machines are combined with image guidance capabilities using cone-beam CT (CBCT) or in-room kV imaging (ExacTrac), a targeting accuracy of 0.5 AE 0.2 mm can be achieved for small intracranial targets. 7 Users of modern linacs as a SRS tool have unprecedented choices of various techniques for the treatment of radiosurgery targets.
A number of groups showed the efficacy of static conformal arc, dynamic conformal arc, or intensity-modulated treatments with MLC delivery to treat various indications including acoustic schwannoma, arteriovenous malformation, meningioma, and metastasis. 8,9 Treatment of trigeminal neuralgia with linac SRS appears to be limited to the use of arcs with circular shaped collimators. 10 To our knowledge, there is no study of MLC-based SRS for trigeminal neuralgia reported in the literature. The purpose of this paper is to demonstrate the validity and reliability of HDMLC SRS delivery for trigeminal cases. In this paper we report results for: (a) verification of MLC beam shapes, (b) total dose delivered to representative treatment volume (a plastic scintillator), and (c) the shape of the cumulative dose distribution (via film).

2.A | Case details
Nine clinical trigeminal neuralgia treatment plans were selected for verification testing. These consisted of a mix of left-sided and rightsided cases, including one patient with bilateral target volumes. The prescribed dose ranged from 6000 cGy for the bilateral case up to 8750-9375 cGy for all other cases. Target volumes ranged in size from 0.026 to 0.068 cc (mean value 0.041 cc) with per-plan equivalent square field size ranges of 0.59-0.70 cm (mean value 0.67 cm).
For each plan, the minimum per-field equivalent square field size was restricted to 0.50 cm, the smallest field size for which output factors and beam profiles were directly measured during TPS commissioning. An example of a trigeminal neuralgia target volume and associated treatment field distribution and isodose distribution near the mean of the group is shown in Fig. 1.
All treatment plans utilized the 6 MV FFF treatment beam and were planned for delivery on a Varian Truebeam STx linear accelerator (Varian Medical Systems, Inc., Palo Alto, CA) equipped with HDMLC (0.25 cm leaf width projection at isocenter) and Exac-Trac with 6D robotic couch (Brainlab AG, Munich, Germany). Plans were delivered with 18-21 static conformal fields (Table 1), depending on the patient geometry and required dose fall-off. All plans were remapped for phantom delivery from clinically accepted patient treatment plans. For this study, all identified cases had an absolute dose measurement performed at isocenter along with aggregated analysis of trajectory log files. Four cases representing the range of field sizes observed were also selected for film dosimetry verification. detector. The scintillating fiber of the W1 has a diameter of 1 mm and a length of 3 mm and is minimally perturbing to the incident radiation field due to its near water-equivalent construction. 11 The W1 was cross-calibrated with an ADCL-calibrated farmer chamber (Exradin A12, Standard Imaging, Inc.).

2.B | Mechanical delivery precision
The BRUNO phantom underwent CT simulation in which a Brainlab mask was constructed for reproducibility of phantom positioning. Axial slices through the entire length of the phantom were acquired at 1 mm thickness for patient plan mapping while 0.5 mm slices were acquired through the extent of the W1 block insert. The two scans were registered within the iPlan treatment planning system and the 0.5 mm images were used for accurate geometric reconstruction of the W1 to facilitate accurate density modeling and dose prediction (Fig. 2).
Clinical treatment plans were mapped to the BRUNO dataset with isocenter placed in the centroid of the scintillating fiber. All table, collimator, and gantry angles were left intact from the clinical treatment plan. After 3D dose computation, plan-specific adjustments to isocenter position were introduced to best center the W1 within the dose cloud such that variation in dose across the fiber was minimized. The magnitude of these isocenter adjustments ranged from 0.0 to 0.8 mm. Dose was recomputed following these fine adjustments, and mean dose to the scintillating fiber structure was recorded as the planned dose value for comparison against measurement.
The scintillator response used in conjunction with the SuperMAX two-channel electrometer (Standard Imaging, Inc.) was calibrated immediately prior to performance of these verification measurements using the methodology recommended by the vendor, which is based on work by a number of researchers. [12][13][14][15] The primary goal of this calibration is to correct for Cerenkov emissions created proportional to the irradiated length of the optical fiber. The calibration process consists of performing large-field irradiations of the W1 with minimum and maximum lengths of the optical fiber within the radiation field and computing a Cerenkov light ratio (CLR) correction according to eq. (1): SC1 and SC2 refer to the charge reading from the blue and green spectral regions of the photodiode while the subscripts MIN and MAX refer to the amount of optical fiber within the radiation field. The gain of the scintillator is further adjusted such that the electrometer reads out directly to dose by application of eq. (2): The additional subscript "10" in the above equation refers to the fact that the calibration and absolute dose is based on the delivery of a known dose using a 10 9 10 cm field in the minimum fiber configuration. This calibration formalism is applied in general to dose measurements according to eq. (3): For treatment delivery the BRUNO phantom was immobilized using the Brainlab mask system constructed during CT simulation.
Six-dimensional CBCT-based corrections to the phantom position were implemented using the ExacTrac robotic couch in order to match the phantom position with the CT dataset used in the verification plan, including the previously mentioned sub-millimeter adjustments to isocenter position.
All treatment fields were delivered at the clinical gantry, collimator, and couch rotational positions. For the W1 measurements there were no additional corrections to the phantom position made after couch rotations, such as through the use of the ExacTrac kV x-ray system. Couch walkout for this particular treatment machine is tested on a monthly basis and has a value of less than 0.6 mm across the full range of couch motion, expressed as the maximum distance between beam center intersections for a couch star-shot film.
Per-field and per-plan doses were tabulated and compared to iPlan calculation for each of the 178 treatment fields spread across the nine patient plans.

Film measurement
Four cases spanning the range of mean field sizes were selected for film analysis. The film insert contains a central low-contrast sphere which was contoured and used for isocenter placement of the mapped treatment group within the iPlan software. Prior to dose calculation the prescription was scaled down to a maximum isocenter dose of 500 cGy in order to facilitate more accurate film calibration over a narrower range. Two-dimensional dose distributions centered on isocenter were exported in DICOM-RT format through the transverse, coronal, and sagittal planes.
Prior to treatment delivery the BRUNO phantom with film insert was positioned using the ExacTrac kV x-ray system. Imaging was repeated after each couch rotation to ensure that couch walkout did not appreciably affect the phantom position. Using a technique of 80 kVp and 10 mAs per image, the total imaging dose contribution according to previous measurements is estimated at approximately 1 mGy, and can safely be ignored. 20 Positioning tolerance with the ExacTrac x-ray system was set to 0.5 mm and 0.5°, and any verification imaging that suggested corrections larger than these values resulted in 6D corrections being applied followed by repeat verification imaging. Each treatment field was delivered at the clinical couch, collimator, and gantry positions and the process was performed a total of three times, once for each film orientation.
Scanned films were read into RIT for analysis, and the dose calibration curve was applied. A 5 9 5 median filter was applied to the scanned images for noise reduction. The FWHM of the dose distribution was determined in all three principal axes from one-dimensional profiles passing through the maximum dose point of each film.
The exported dose planes were read into RIT as reference images, and the FWHM of the planned dose was likewise obtained. The planned and measured dose planes were co-registered, and gamma analysis of both the 2D planes and 1D profiles through isocenter using a global 3%/1 mm criteria was performed. 2D isodose overlays were generated and inspected visually to further establish agreement between the planned and delivered dose distributions.

3.A | Mechanical delivery precision
The distribution of MLC leaf deviations from their planned positions is shown in Fig. 3 3.B | End-to-end testing

3.B.1 | Absolute dose verification
Composite absolute dose results for the nine clinical cases showed a percentage difference from treatment planning system prediction of À2.8% to +1.1% with a mean percentage difference of À0.4% (Table 1).
On a per-field level the mean discrepancy was À0.5% while the mean discrepancy magnitude was 2.8%. Studies characterizing the scintillator detector used in this work have found an overall uncertainty between 1.0% and 1.7%. 11 The larger dose discrepancies between calculated and measured dose are thus not simply due to accuracy of the measurement device, and likely originate from a combination of (a) treatment planning system algorithm accuracy and (b) accuracy of detector placement for these very small fields. Although the support structures were included in the calculation surface model of the phantom within the iPlan software, this seems to indicate an opportunity for improved support density modeling.
Any such improvements to modeling of patient supports will be particular the treatment planning system being used, and it is incumbent on individual institutions to evaluate their particular setup. Another possibility is that the CBCT-based rotational corrections applied to the BRUNO phantom prior to measurement resulted in an inaccurate geometric representation of the support structures within the plan.
That is, the 6D corrections brought the phantom geometry into alignment with the plan but resulted in the support structures no longer having the plan-assumed geometry. Treatment fields which were calculated to traverse the support structures may not have done so (and vice versa) or they may have traversed a shorter or longer path-length through those structures. In practice, we attempt to minimize the use of beams that pass through the patient support structures for this reason.

FWHM (mm)
Lt./rt.  Measured and planned isodoses, dose profiles, and gamma analyses for one representative film are shown in Fig. 5. The largest difference tended to occur either at the field center or at the outer edges of the penumbra between the 10% and 30% isodose lines.
The tendency to measure cold near the isocenter and hot in the 2D gamma analysis was performed using a 3%/1 mm passing criteria (Fig. 5) 26 Huang et al. 7  In conjunction with a quality-controlled treatment delivery methodology, an appropriately commissioned treatment planning system can be used for accurate and clinically appropriate design of trigeminal neuralgia treatment plans utilizing HD-MLC.

CONFLI CT OF INTEREST
No conflicts of interest to disclose.