Quantifying false positional corrections due to facial motion using SGRT with open‐face Masks

Abstract Purpose Studies have evaluated the viability of using open‐face masks as an immobilization technique to treat intracranial and head and neck cancers. This method offers less stress to the patient with comparable accuracy to closed‐face masks. Open‐face masks permit implementation of surface guided radiation therapy (SGRT) to assist in positioning and motion management. Research suggests that changes in patient facial expressions may influence the SGRT system to generate false positional corrections. This study aims to quantify these errors produced by the SGRT system due to face motion. Methods Ten human subjects were immobilized using open‐face masks. Four discrete SGRT regions of interest (ROIs) were analyzed based on anatomical features to simulate different mask openings. The largest ROI was lateral to the cheeks, superior to the eyebrows, and inferior to the mouth. The smallest ROI included only the eyes and bridge of the nose. Subjects were asked to open and close their eyes and simulate fear and annoyance and peak isocenter shifts were recorded. This was performed in both standard and SRS specific resolutions with the C‐RAD Catalyst HD system. Results All four ROIs analyzed in SRS and Standard resolutions demonstrated an average deviation of 0.3 ± 0.3 mm for eyes closed and 0.4 ± 0.4 mm shift for eyes open, and 0.3 ± 0.3 mm for eyes closed and 0.8 ± 0.9 mm shift for eyes open. The average deviation observed due to changing facial expressions was 1.4 ± 0.9 mm for SRS specific and 1.6 ± 1.6 mm for standard resolution. Conclusion The SGRT system can generate false positional corrections for face motion and this is amplified at lower resolutions and smaller ROIs. These errors should be considered in the overall tolerances and treatment plan when using open‐face masks with SGRT and may warrant additional radiographic imaging.


| INTRODUCTION
Radiotherapy treatment of tumors in regions of the head and neck require reproducible and accurate techniques. Some stereotactic radiosurgery (SRS) treatments have shifted from the exclusive method of invasive frame-based treatments to include an additional method using frameless, moldable masks for treatment delivery. 1 Frame-based SRS consists of a rigid frame screwed into a patient's skull for a single fraction of radiation, 2 while frameless masks consist of a thermoplastic material that is molded to the patient's face at the time of CT simulation. The clinical goal has focused on improving patient comfort without losing treatment accuracy. The frameless masks have been found to be a practical and reproducible method for image-guided SRS treatments providing spatial accuracy comparable to frame-based treatments. 3 The drawback of the thermoplastic masks is its enclosed method of immobilization that can cause high levels of stress as patients are forced to keep their eyes and mouth closed throughout the treatment process 4 . The immobilization process can cause anxiety severe enough to disrupt the session, 5 and patients may experience feelings of fear, anger, or depression. This discomfort has resulted in the introduction of an alternative immobilization method that partially exposes a patient's face for intracranial 6 and head and neck 4 radiotherapy cancer treatments. As a result, claustrophobic patients immobilized with this open-face style mask have reported less distress during treatment delivery. 4 Positional uncertainty can negatively impact the accuracy of radiotherapy treatments, so understanding head motion for this immobilization method is important for determining clinical treatment margins. 4 Head motion, for patients immobilized with closed-face thermoplastic masks, has generally been characterized based on x-ray images before and after treatment. 7,8 Open-face masks allow for realtime motion monitoring when coupled with surface guided radiation therapy (SGRT) systems both during patient set up 9 and during treatment delivery. 6,10,11 Without the use of radiation, SGRT systems use optical imaging, to generate 3D maps of a patient's surface. A registration algorithm then compares the live image to a baseline reference image to monitor deviations from original treatment position. Direct imaging of the skin has been considered a more accurate method for motion management as moldable masks may become loose, making small head motion undetectable. 4 Measurements with phantoms have shown that SGRT is a suitable and reproducible option for intra-fraction radiosurgery localization. 12,13 An open-face mask that exposes the entire face has been recommended for anthropomorphic head phantom and patient setup with SGRT. 14 Several manufacturers produce open face masks, some expose only the patient's eyes, while others expose the eyes, mouth, and cheeks.
Recent research has aimed to characterize the immobilization performance of using open face masks with SGRT. One study reported that use of a small mask opening or SGRT region of interest (ROI) may generate an apparent shift or false positive with the SGRT system from changes in facial expression (ex. smiling). 4 This apparent shift demonstrates a potential concern that clinics should consider when using SGRT with frameless open face masks. The primary goal of this work was to demonstrate a previously unexplored area of potential error in SGRT tracking. Our study aims to further quantify false positional shift corrections generated by the SGRT system due to face motion by evaluating multiple SGRT ROIs using two spatial resolution settings. It is our hypothesis that the SGRT system will generate false positional corrections for face motion and they will be amplified at a lower spatial resolution setting and at smaller ROIs. To measure this, human subjects were immobilized using open-face masks and discrete SGRT ROI were monitored. This experiment recorded positional corrections generated by the SGRT system as human subjects opened and closed their eyes and changed facial expressions to simulate emotions. These methods were performed using two different camera resolutions using the C-RAD Catalyst HD system.

2.A | C-RAD Catalyst HD
The C-RAD Catalyst HD (C-RAD, Uppsala, Sweden) uses three ceiling scanner units (see Fig. 1), consisting of a light projector and CCD camera, positioned equidistant above the radiotherapy treatment couch. This system uses a dose-free triangulation method 15  spatial resolutions: standard and SRS. Standard resolution incorporates a non-rigid algorithm that is beneficial for non-rigid treatment sites in regions of the breast or extremities while SRS incorporates a higher resolution with more calculation points along with a more robust algorithm to accommodate for open-face masks. 16 The camera has a temporal resolution of 200 frames/sec and its sensitivity can be adjusted with two parameters, Integration time (µs) and Gain (%), to account for difference in reflection produced by different skin tones. 17

2.B | Methods
With Institutional Review Board (IRB) approval, ten healthy human subjects (five female and five male) were used to quantify potential positional deviations due to eye movement and facial expressions with the use of an SGRT system. Individuals were immobilized using We would expect similar efficacy when using a bright colored mask on a darker skin tone however to reduce variables in this study, a common mask surface color was used.
To be consistent, individuals were positioned with the Catalyst HD using the SRS resolution (~Time:10,000µs, Gain: 0%), so that the center of their head or "isocenter" was at C-RAD's central axis. Camera settings were adjusted to allow for maximum optimization based  (see Fig. 4). Since a cancer diagnosis and course of treatment can induce a range of different emotions, 18 we asked volunteers to simulate fear and annoyance. We considered these to be common emotions for patients as they may be fearful of receiving a radiation treatment or annoyed by the treatment process. Each subject was asked to "close eyes (C1), open eyes, close eyes (C2), express fear, close eyes (C3), express annoyance, and close eyes (C4)." Human subjects were asked to close their eyes between each activity to determine whether patient positioning returned to a steady baseline.
Peak translational (Vert, Long, Lat) and angular (Rot, Roll, Pitch) shifts generated by the SGRT system were manually recorded for each task and the total vector shift or total deviation from the isocenter was calculated based on translational shifts using [Eq. (1)].
To evaluate the efficacy of ROI size and impact of face motion, the average value and standard deviation for discrete peak values of a single task performed were calculated across similar ROI sizes (ROI 1, ROI2, etc.) for all subjects (1, 2, 3, etc.). We calculated a statistical deviation (SD), as seen in Tables 1 and 2         and improve preparedness for treatment. 19 Therapists should ask patients to remain relaxed with their eyes closed throughout the duration of treatment simulation and delivery.

3.B | Standard resolution
T A B L E 2 Comparison of values for the positional deviations generated by the SGRT system in Standard resolution for eye motion and emotions. Data is presented for ten subjects wearing open-face masks using four discrete ROIs with C-RAD's Catalyst HD