Assessment of the use of different imaging and delivery techniques for cranial treatments on the Halcyon linac

Abstract Purpose In this work, we investigated the effect on the workflow and setup accuracy of using surface guided radiation therapy (SGRT) for patient setup, megavoltage cone beam CT (MVCBCT) or kilovoltage cone beam CT (kVCBCT) for imaging and fixed IMRT or volumetric‐modulated arc therapy (VMAT) for treatment delivery with the Halcyon linac. Methods We performed a retrospective investigation of 272 treatment fractions, using three different workflows. The first and second workflows used MVCBCT and fixed IMRT for imaging and treatment delivery, and the second one also used SGRT for patient setup. The third workflow used SGRT for setup, kVCBCT for imaging and VMAT for delivery. Workflows were evaluated by comparing the number of fractions requiring repeated imaging acquisitions and the time required for setup, imaging and treatment delivery. Setup position accuracy was assessed by comparing the daily kV‐ or MV‐ CBCT with the planning CT and measuring the residual rotational errors for pitch, yaw and roll angles. Results Without the use of SGRT, the imaging fields were delivered more than once on 11.1% of the fractions, while re‐imaging was necessary in 5.5% of the fractions using SGRT. The total treatment time, including setup, imaging, and delivery, for the three workflows was 531 ± 157 s, 503 ± 130 s and 457 ± 91 s, respectively. A statistically significant difference was observed when comparing the third workflow with the first two. The total residual rotational errors were 1.96 ± 1.29°, 1.28 ± 0.67° and 1.22 ± 0.76° and statistically significant differences were observed when comparing workflows with and without SGRT. Conclusions The use of SGRT allowed for a reduction of re‐imaging during patient setup and improved patient position accuracy by reducing residual rotational errors. A reduction in treatment time using kVCBCT with SGRT was observed. The most efficient workflow was the one including kVCBCT and SGRT for setup and VMAT for delivery.


| INTRODUCTION
In recent years there has been a growing interest in the assessment of the performance and quality of the treatment delivery process for external beam radiotherapy, 1-4 favoring a more efficient and safer clinical practice. In particular, time-efficient workflows during radiotherapy treatments allow for a higher clinical throughput, improved patient comfort and reduced costs per treatments by reducing the machine and staff hours. Achieving such efficient workflows require a thorough assessment and optimization of each stage of the treatment process, including patient setup, image-guided evaluation and delivery of the treatment fields. Halcyon speeds are up to four and two times higher during the imaging and dose delivery stages, respectively. Such increased rotational gantry speeds, combined with a higher multileaf collimator (MLC) leaf speed allow for faster treatments. The tradeoff between plan quality and time efficiency in Halcyon has been investigated for some sites as prostate 5 and head and neck. 6 In both cases, the plan quality with Halcyon was maintained but the imaging and dose delivery times were reduced as compared with treatments using a C-arm linac.
In a typical treatment with Halcyon, the patient is aligned to the virtual isocenter, a reference point outside the bore, and then, the couch is moved into the bore to the treatment isocenter. Image registration to the CT is performed using a 2D-3D image registration when the orthogonal pairs are used, or a 3D-3D registration when the CBCT technique is used, to calculate and apply the shifts between the CT and daily images. The couch in Halcyon only allows for translational shifts, and, if large rotations are observed on the initial images, the patient needs to be manually re-aligned and additional images are taken. Several authors have reported the dosimetric consequences of rotational setup errors. 7-10 Peng et al. reported changes up to 11% on the clincal target volume (CTV) coverage for cranial stereotactic radiosurgery (SRS), 7 and Briscoe showed that setup accuracy is especially critical for plans treating multiple metastases with a single isocenter. 10 Initial setup accuracy is relevant, not only to ensure accurate dose delivery, but also to improve the workflow and prevent excess imaging fields. The time for correcting the setup and taking new images is added to the total treatment time, thus decreasing the efficiency of the process and worsening the patient experience as he or she has to spend more time on the couch. Surface Guided Radiation Therapy (SGRT) provides an option for improved patient setup and real-time monitoring and has been used previously for intracranial [11][12][13][14] and extracranial regions. [15][16][17][18][19][20][21] SGRT is a nonradiographic localization technique that uses a system of 3D cameras to detect a light pattern projected onto the patient and use the information to reconstruct a 3D surface image. The generated 3D image can be compared real time to the body contour from the CT scan at simulation and assist with patient setup and tracking.
The purpose of this work was to evaluate the optimal imaging modalities and dose delivery techniques, for cranial treatments using the Halcyon linac considering workflow efficiency and setup accuracy.

| METHODS AND MATERIALS
At our institution, non-SRS cranial treatments are performed using the Halcyon linac. For this set of patients, all undergoing standard fractionations and with doses per fraction ranging from 1.8 Gy to 3.0 Gy, the planning simulation is performed with a General Electric computed tomography G scanner using a slice thickness of 2.5 mm and immobilization is achieved by using a head rest and a full mask.
In this work, we compared retrospectively the efficiency and setup accuracy of three workflows using different setup, imaging and dose delivery modalities.  Figure 1 shows the workflow at our institution for intracranial treatments before the implementation of SGRT.

2.A | SGRT implementation
AligntRT ® (Vision RT Ltd, London UK) is a video-based 3D camera system used for SGRT. Figure 2 shows the configuration for AligntRT at the Halcyon vault on our clinic, consisting of a two-camera system, located at the ceiling and aligned to the virtual isocenter defined by the lasers out of the bore. Daily calibration is performed by using a plate with a circle pattern. For each camera, the system detects reference markers, whose locations are refined by the user, and determines the isocenter. Unlike conventional implementations of AlignRT, the two-camera system on the Halcyon linac cannot be calibrated to the treatment isocenter due to blockage by the gantry bore. Therefore, the calibration is performed using the virtual isocenter. The coincidence between the virtual and treatment isocenter is measured by an integrated self-check tool, the machine performance check In order to incorporate the use of SGRT, the workflow was modified adding one step for loading the patient and another to setup the patient using SGRT (Fig. 4). Currently, there is no surface tracking after the thermoplastic mask is placed.

2.B | Phantom-based accuracy verification
The accuracy of the method used in this work to measure the residual rotational errors was evaluated using the AlignRT isocenter calibration phantom, which has five embedded ceramic spheres and is mounted on a base with three leveling screws. The phantom was scanned with a GE CT scanner using a head protocol with 1.25 mm slices and a surface image was generated using the Eclipse™ treat-

2.C | Data acquisition
Approval for the study was granted by the University of California San Diego Institutional Review Boards, project #181861XL. Table 1 shows the demographics for the fractions investigated in this work.
Plans investigated included treatment sites such as orbits, head and neck, glioblastoma multiple (GBM) and brain, with treatment volumes ranging from 9.5 cm 3 to 456 cm 3 . Data for 272 fractions, distributed among 15 patients, were analyzed. Five metrics were used to evaluate the impact over the workflow and dose delivery accuracy of different setup, imaging and delivery modalities: • Percent of fractions with additional imaging. The number of images that were acquired in addition to the planned images because the patient had to be manually repositioned after the initial setup. We compared 117 fractions without SGRT with 155 fractions using SGRT.   All data were analyzed calculating the mean, median, maximum and quartile distributions. The nonparametric Mann-Whitney test was used to establish the statistical significance of the results at P < 0.01 level.

3.B | Time for setup, imaging and treatment fields
Without the use of SGRT, additional imaging fields were necessary in 11.1% of the fractions. Once SGRT was incorporated into the workflow, the percentage of fractions requiring more than one imaging field was reduced to 5.5% (Table 2).     requires further investigation and is expected to be more relevant when the treated area is closer to organs at risk.
In this work we have shown the advantages of using SGRT for initial setup of cranial treatments using the Halcyon linac.
However there are other sites that could also benefit from this technique such as head and neck, extremities, breast and prostate as has been shown for C-arm linacs. [15][16][17][18][19][20][21] We identified two limitations of the current SGRT setup. First, the use of full masks precludes surface tracking after mask placement. It is possible that some of the patients move during mask placement and such deviations are not detected prior to the setup fields. A second limitation is that the cameras are calibrated to the virtual isocenter and do not monitor the patient during the treatment. Future developments on SGRT capable to monitor the patient at the radiation isocenter will allow monitoring intrafractional motion by real-time tracking.

| CONCLUSIONS
We have shown that the most efficient workflow for patients treated in the cranial region using Halcyon is the one that uses SGRT for patient setup, kVCBCT for imaging and VMAT for delivery. The use of SGRT allows for a reduction over the number of additional imaging fields on the patient setup, and a more accurate dose delivery by means of a significant reduction of the residual rotational error.

CONF LICT OF I NTEREST
UC San Diego received in-kind funding from Vision RT Ltd in partial support of this research project.