Surface guided radiotherapy (SGRT) improves breast cancer patient setup accuracy

Abstract Purpose The purpose of the study was to investigate if surface guided radiotherapy (SGRT) can decrease setup deviations for tangential and locoregional breast cancer patients compared to conventional laser‐based setup (LBS). Materials and Methods Both tangential (63 patients) and locoregional (76 patients) breast cancer patients were enrolled in this study. For LBS, the patients were positioned by aligning skin markers to the room lasers. For the surface based setup (SBS), an optical surface scanning system was used for daily setup using both single and three camera systems. To compare the two setup methods, the patient position was evaluated using verification imaging (field images or orthogonal images). Results For both tangential and locoregional treatments, SBS decreased the setup deviation significantly compared to LBS (P < 0.01). For patients receiving tangential treatment, 95% of the treatment sessions were within the clinical tolerance of ≤ 4 mm in any direction (lateral, longitudinal or vertical) using SBS, compared to 84% for LBS. Corresponding values for patients receiving locoregional treatment were 70% and 54% for SBS and LBS, respectively. No significant difference was observed comparing the setup result using a single camera system or a three camera system. Conclusions Conventional laser‐based setup can with advantage be replaced by surface based setup. Daily SGRT improves patient setup without additional imaging dose to breast cancer patients regardless if a single or three camera system was used.

radiotherapy, one breast cancer death can be avoided. 1 There is no effective method to find microscopic disease after breast conserving surgery and therefor radiotherapy is still considered to be important for the cure of breast cancer. Radiotherapy for breast cancer treatment uses a three-dimensional computed tomography (3DCT)based treatment planning which enables a high local selectivity for the dose distribution; the target tissue is irradiated while the normal tissue is spared. The treatment planning system (TPS) ensures a high accuracy in the dose deposition which requires high accuracy in daily patient setup. Breast cancer patients have a long expected survival and it is of importance to reduce interfractional setup errors to avoid excessive irradiation that can cause toxicity in normal healthy tissue. The organs at risk (OAR) are primarily the lung and the heart. Hence, complications such as radiation pneumonitis and cardiac mortality have been shown to positively correlate with the volume irradiated. 2,3 Setup verification imaging strategies, generally classified as either online or offline, are used to ensure that systematic and randomized setup deviations are minimized throughout treatment. The drawback is that both strategies are associated with a risk for second malignancies due to imaging dose. 4 The online strategy implies daily imaging before treatment with a preset threshold for deviations. Laaksomaa et al., recommended daily online image guidance due to large random interfractional variation in patient posture. 5 This strategy is time-consuming and contributes imaging dose to the patient throughout treatment.
Having in mind the increased radiation dose due to imaging, the ALARA principle and the fact that the survival of breast cancer patients is expected to be long, an accurate nondose-contributing setup system is warranted. The offline strategy requires frequent imaging in the beginning of the treatment course. The result is statistically analyzed for the systematic and random components of the deviation in the patient position. The systematic deviation is compensated for by a couch shift for the following treatment sessions. 6 The random deviation is mainly due to the inaccuracy in laser aligned setup, which is commonly used for daily setup. The patients are aligned according to landmarks on the skin and room lasers. 6 An alternative approach is to use surface guided radiotherapy (SGRT), which uses a three-dimensional (3D) model of the skin surface for positioning and monitoring. The optical surface scanning (OSS) system compares a 3D model of the patient's external surface extracted from the TPS with a live scan of the surface while the patient is positioned on the treatment couch ( Fig. 1). Surface based setup (SBS) increases the patient setup information compared to laser-based setup (LBS), by using the entire patient skin surface instead of only three skin marks. Several OSS systems have shown a high correlation with verification imaging results. 7-9 Also, Chang et al. have in a study with 23 patients shown that SGRT has a high correlation to the lumpectomy cavity defined by surgical clips for breast cancer patients receiving accelerated partial breast irradiation. 10 The OSS system Catalyst TM (C-rad Positioning AB, Uppsala, Sweden) has been evaluated in this study. This OSS system is unique because it uses a deformable algorithm to calculate the isocenter position.
The principle behind the deformable registration in depth scans is described by Hao Li et al. 11 Recently published results showed that patient setup using the deformable algorithm of the Catalyst TM system was superior to LBS for breasts with nodal involvement in TomoTherapy (Accuray, Sunnyvale, CA). 12 13 However, comparison between tangential and locoregional treatments and single vs. three camera systems has to our knowledge not been investigated.
Tangential and locoregional treatments, and also, single and three camera systems result in different surface coverage which motivates an investigation of how the setup accuracy is affected.
The aim of this study was to retrospectively compare LBS with SBS using the OSS system Catalyst TM for both tangential and locoregional breast cancer patients using single and three camera systems.

2.A | Ethical consideration and consent
The use of the radiotherapy database for retrospective research has been approved by the Regional Ethical Review Board in Lund (No. 2013/742).

2.D | Treatment plans
The treatment prescription was 50 Gy in 25 fractions or 42.6 Gy in 15 fractions, normalized to the PTV mean or median dose. In the TPS 3D conformal treatment plans were created for all patients. For the tangential treatments, two opposing 6 MV tangential fields to cover the breast tissue was used. Also, a supplementary field and/or wedges were used for dose homogenization purposes. The isocenter position was placed central in the breast tissue. For the locoregional treatments, opposing tangential fields were used to cover the location of the breast tissue. To complement the tangential fields in order to achieve homogeneous dose a various number of supplementary fields of 6 or 10 MV were used. The number of fields used depended on target size and patient anatomy. The locoregional axillary lymph nodes were covered with a 6 MV anterior-posterior (AP) field and a 10 MV posterior-anterior (PA) field. Also, a supplementary 10 MV PA field was used while shielding of the lung tissue.
The total number of fields used for the locoregional treatments ranged between six and nine. For mastectomy patients, a 0.5-cm thick and 6-cm wide bolus (Superflab, Mick Radio-Nuclear Instruments, Inc. An Eckert & Ziegler BEBIG Company) was placed over the operation scar. For locoregional treatment, the treatment isocenter was positioned in the junction between the tangential and AP-PA fields.
2.E | Surface guided radiotherapy with a deformable algorithm for isocenter calculation The OSS systems were ceiling-mounted in nine treatment rooms, as either a single camera or a three camera configuration. The three camera configuration provides a 360°surface coverage of the patient, due to a 120°installation angle between the systems. For erence and live surface match the system carries out a depth calculation of the isocenter position using a deformable algorithm (Fig. 1). 16 Thus, the calculated isocenter shift depends on the daily patient setup. For the isocenter calculation, the full patient surface coverage of the thorax was used, however, surface close to isocenter is weighted higher in the calculation than distant surface. Also, anatomical deformations that can occur during the course of radiotherapy were automatically handled by the algorithm. For each KÜGELE ET AL. patient, the scanning parameters were adjusted individually in the Catalyst TM software to obtain optimal image quality, minimize camera shadowing, and over or under exposed images.

2.F | Patient setup protocol
At the treatment machine, all patients were initially positioned by laser alignment to a 3-point based tattoo setup.
For the patients positioned using LBS, the calculated shift from the reference point to the isocenter position was manually carried out the first treatment session and the isocenter position was drawn onto the patient's skin using a marker pen. Verification images were acquired, according to a No Action Level (NAL) offline strategy. 17 The systematic deviation was estimated after the first three treatment sessions and the setup was corrected for the remaining treatment sessions. To carry out a fair comparison between the LBS and SBS setup strategy, the setup data from the three first treatment session for the patients positioned using LBS were excluded.
For the patients positioned using SBS, the couch was shifted to the treatment position and the correction for posture was performed using the color map with a tolerance of 5 mm (Fig. 2). The effect of the free breathing motion was minimized by using a floating mean value calculation over 4 s for the live image. Once the posture was within the surface tolerance, the therapists manually shifted the couch to correct for the isocenter deviation. The patient setup result was saved by the therapists inside the treatment room and a residual isocenter deviation ≤2 mm, and rotations ≤ 3°were accepted in this study.
For both the tangential and locoregional treatments, each patient was positioned using either SBS or LBS and the position was verified using onboard imaging at the linac. Different patient anatomies were included for all four groups. The shifts in lateral (lat), longitudinal (lng) and vertical (vrt) direction, respectively, and the total vector off- ), were evaluated. For SBS, the variation in patient anatomies caused more or less camera shadowing in the live image (Fig. 3). Students t-test for two independent mean for the lat, lng, and vrt directions, with a statistical level of significance α = 0.01.

2.F.2 | Locoregional treatment
Three patient groups of totally 76 patients were enrolled in this study. For SBS, 43 patients were included; 22 patients positioned using a three camera system and 21 patients positioned using a single camera system. Nineteen of the patients had bolus over the operation scar, and one patient had a 1-cm thick wet towel as a bolus. One patient was excluded due to that the OSS system was not used according to the study protocol. In the LBS group, 34 patients were enrolled. The patient setup was verified with orthogonal kilovolt (kV) or megavolt (MV) images. The anatomical landmarks used were the clavicular bone position, the lung edge, and sternum.
In total, 632 verification images were evaluated. For comparison a two-sided Wilcoxon sum rank test was carried out for the vector offset and Students t-test for two independent mean for the translational directions with a statistical level of significance α = 0.01.
A two-sided Wilcoxon sum rank test was carried out to investigate if there were any statistical significant difference between the single and three camera system with a significance level of 0.01.

3.B | Locoregional treatment
The median vector offset was 4.7 (0-18.7 mm) and 4.0 (range: 0−13.5 mm) for LBS and SBS, respectively (p < 0.01). For LBS, the mean value (±1 SD) was 0.1 ± 3.4 mm, 0.1 ± 3.3 mm, 0.7 ± 3.1 mm in lat, lng, and vrt directions, respectively. For SBS, the mean value (±1 SD) was −0.5 ± 2.8, −0.1 ± 2.8, −0.3 ± 2.9 mm in lat, lng, and vrt directions, respectively. The result was statistically significant for lat and vrt directions (P < 0.01). The cumulative probability for positioning a patient within a spatial vector of 5 mm from isocenter was 55% for LBS and 67% for SBS (Fig. 5). For 90% of the treatment sessions, the spatial vector was within 9.1 and 7.6 mm for LBS and SBS, respectively. For LBS and SBS, 54% and 70% of all treatment sessions were within the clinical tolerance of ≤4 mm in all three directions (lat, lng, vrt), respectively (Fig. 5). A small but not significant difference was observed (P = 0.02) for the vector offset, comparing the single camera system with the three camera system for SBS (Fig. 6).

| DISCUSSION
For both the patient groups receiving tangential and locoregional breast cancer treatment, the patient setup was significantly improved using the Catalyst TM system. For locoregional treatments, the clinical criteria (≤4 mm) were fulfilled for 16% more treatment sessions using SBS compared to LBS. The corresponding improvement for tangential treatments was 11%. This could potentially lead to a reduced amount of verification imaging in the clinic. Also, the standard setup deviation for patients receiving tangential treatment was where the bolus often is positioned, vital information is lost. For a single camera system and a locoregional patient with bolus, the surface that was covered was the lower parts of the thorax, arm, and chin. Since the arm and chin are not optimal anatomical structures to use for patient setup, this contributes to inaccuracy in this study.
For the three camera system, better surface coverage over the treatment area was observed, contributing to a more accurate patient setup. The single camera system was installed in the ceiling by the foot end of the couch. The reconstructed surface depended on how much of the patient surface the camera was able to detect. For tangential treatments, a breastboard pitch of 7.5°was used which favored the Catalyst TM camera, hence, the patient was tilted toward the camera. For patients receiving locoregional treatments, a breastboard pitch of 0°was used and in combination with a cranial isocenter, important surface above the isocenter was not covered using the single camera system. This loss of surface had a negative impact on the accuracy of the patient setup (Fig. 3). For 17 out of the 76 patients in this study, the mass center of the PTV was used instead of the isocenter for the setup calculation in the Catalyst TM system.
However, since the surface above the calculation point was lost, the setup accuracy was similar to using the calculated isocenter. For the three camera system, for locoregional treatment, and single camera for tangential treatment, the treatment site was well covered, which according to our results, as well as in the study of Chang et. al, leads to accurate positioning. 10 In the time span between the in room SBS and the verification imaging during treatment, patient motion con-

| CONCLUSION
This study showed that surface based setup, using the Catalyst TM system can replace the conventional laser-based setup for tangential and locoregional breast cancer treatments, regardless if a single or a three camera system was used. Additional information of the patient posture was provided using surface based setup compared to laserbased setup, which improved positioning. Daily surface guided radiotherapy for breast cancer patients can thus reduce time and dose associated with verification imaging.

ACKNOWLEDG MENT
Our research is partly financially supported by C-RAD AB, Uppsala, Sweden, however, the authors alone are responsible for the content, analyses, and writing.
F I G . 6. For the breast cancer patients receiving locoregional treatment, a total of 362 verification images were evaluated for patients positioned using SBS at a single camera Catalyst system (181 verification images) and a three camera Catalyst system (181 verification images). The cumulative probability of the vector offset shows a nonsignificant improved patient setup for three camera compared to single camera system (P = 0.02).