Monitoring frequency of intra‐fraction patient motion using the ExacTrac system for LINAC‐based SRS treatments

Abstract Purpose The aim of this study was to investigate the intra‐fractional patient motion using the ExacTrac system in LINAC‐based stereotactic radiosurgery (SRS). Method A retrospective analysis of 104 SRS patients with kilovoltage image‐guided setup (Brainlab ExacTrac) data was performed. Each patient was imaged pre‐treatment, and at two time points during treatment (1st and 2nd mid‐treatment), and bony anatomy of the skull was used to establish setup error at each time point. The datasets included the translational and rotational setup error, as well as the time period between image acquisitions. After each image acquisition, the patient was repositioned using the calculated shift to correct the setup error. Only translational errors were corrected due to the absence of a 6D treatment table. Setup time and directional shift values were analyzed to determine correlation between shift magnitudes as well as time between acquisitions. Results The average magnitude translation was 0.64 ± 0.59 mm, 0.79 ± 0.45 mm, and 0.65 ± 0.35 mm for the pre‐treatment, 1st mid‐treatment, and 2nd mid‐treatment imaging time points. The average time from pre‐treatment image acquisition to 1st mid‐treatment image acquisition was 7.98 ± 0.45 min, from 1st to 2nd mid‐treatment image was 4.87 ± 1.96 min. The greatest translation was 3.64 mm, occurring in the pre‐treatment image. No patient had a 1st or 2nd mid‐treatment image with greater than 2 mm magnitude shifts. Conclusion There was no correlation between patient motion over time, in direction or magnitude, and duration of treatment. The imaging frequency could be reduced to decrease imaging dose and treatment time without significant changes in patient position.


| INTRODUCTION
Frameless stereotactic radiosurgery (SRS) has taken on a significant role in treatment of cranial lesions, including primary and metastatic brain tumors, nerve disorders, and arteriovenous malformations. SRS provides an alternative to surgery, and whole brain radiotherapy (WBRT), or can accompany these treatments to ensure residual tumor cells are eliminated. Due to the high dose, sharp dose gradients, and small margins utilized in SRS, accurate patient positioning is vital to reduction in dose to normal tissue, as well as tumor control. 1 To achieve the required levels of setup accuracy, image guidance and a thermoplastic mask attached to the treatment couch are used in place of an invasive head frame. Previous works have shown that intra-fractional positioning accuracy of mask-based immobilization systems range from 1.59 AE 0.84 mm to 4.7 AE 1.7 mm using a thermoplastic mask and image guidance from Cone-beam CT (CBCT), CT simulation, portal images, and biplanar diagnostic x ray. [2][3][4][5] These positioning errors are still too large for SRS treatments, due to irradiating critical organs during the treatment. A study by Kim et al. mea-sured the intrafraction shift of 16 patients and found the average to be 0.39 mm, however, this small shift resulted in an average variation in maximum dose to organs at risk (OAR) of 7.15%. 6 Image guidance significantly reduces the setup errors, and is essential for accurate delivery of SRS. Multiple systems have been developed for image guidance, including electronic portal imaging devices (EPIDs), stereoscopic kV imaging, CBCT, and MVCT. 7 A study by Ramakrishna et al. investigating intra-fractional motion found that there was less than 1.0 mm discrepancy between frame-based and imageguided at initial setup using a stereoscopic kilovoltage x ray system combined with an infrared position tracking system, and a positioning error of 0.7 mm for image-guided setup. 8 These imaging methods are highly reliant on bony anatomy for alignment due to the inability to distinguish brain metastases. Previous works have determined that the skull is a reliable surrogate for tumor position. 2,9,10 This study investigates the intra-fractional motion during SRS treatment utilizing a thermoplastic mask and repositioning during treatment using ExacTrac stereoscopic kV x ray system based on our institutional imaging protocol.

| ME TH ODS
A total of 104 sequential patients who had undergone single fraction SRS treatment for brain tumors were retrospectively chosen for this study. All patients had been treated on a clinical Linear accelerator (Trilogy, Varian Medical Systems, Palo Alto, CA), with a thermoplastic mask used for patient immobilization, and image guidance using the ExacTrac kV X-ray system and ExacTrac software version 5.5 (Brainlab, Munich, Germany). Thermoplastic masks are from BrainLab, model 41100, and cover from the patient's forehead, to just above the upper lip. No bite block is used for mask positioning. Thermoplastic masks were formed after heating in a water bath, immediately prior to CT-simulation, an image of an example mask is shown in Fig. 1 All patients were imaged three times over the course of treatment, once pre-treatment, and twice during treatment (1st and 2nd mid-treatment). Patients were initially setup using the in-room lasers and infrared markers, then the pre-treatment image was acquired.
After the pre-treatment image, if any shift was required a second x ray was acquired for shift verification. For mid-treatment images, a verification x ray was acquired for shifts > 2.0 mm in magnitude.
Mid-treatment images occurred between treatment fields and couch rotations, in one of two configurations shown in Fig. 2(a). Figure 2

| RESULTS
The average shifts for all directions were less than or equal to 0.15 mm over all imaging time points, however, the magnitude translations were 0.64 AE 0.59 mm, 0.79 AE 0.45 mm, and 0.65 AE 0.35 mm for pre-treatment, 1st mid-treatment, and 2nd midtreatment image, respectively, as shown in Table 1. Table 2        in frameless SRS patients. Although the dosimetric impact of these imaging sets is low, the process adds to the overall patient time on table, introduces the potential for error, and it is always advantageous to reduce imaging dose to the patient. Reducing imaging frequency would reduce the required treatment time, which is a concern due to the use of a thermoplastic mask covering the patient's face during treatment that can cause nervousness or discomfort. At our institution with our current practice, it is reasonable to reduce imaging frequency to one pre-treatment image, and one mid-treatment image, occurring approximately halfway through treatment delivery.

CONF LICT OF I NTEREST
No author has any conflicts of interest.