Background
In radiotherapy, the intrafractional motion of tumors can become a substantial uncertainty. For example, the motion of lung tumors may reach a peak-to-peak amplitude of up to 24 mm [
1] or even 38 mm [
2]. Liver tumors may reach 34 mm [
3]. There are several approaches to mitigate this uncertainty [
4], the conventional approach being the target volume expansion to cover all the possible positions of the tumor. While this approach ensures the dose coverage of the tumor, it increases the dose to surrounding healthy tissue. Alternative approaches are gating and dynamic tumor tracking techniques. In gated treatment, the tumor motion or its surrogate is monitored and the radiation beam is only switched on e.g. during a specific respiration phase. Dynamic tumor tracking techniques compensate the tumor motion continuously by moving the beam (CyberKnife [
5], Vero [
6]), modifying the beam (MLC tracking [
7]) or counter-steering the patient with the robotic couch, which is called couch-tracking.
The advantages of couch-tracking are 1) it can be performed on a conventional linear accelerator as opposed to beam tracking and 2) it does not disturb the beam as opposed to MLC tracking. However, couch-tracking may influence the respiration of the patient or may induce stress or motion sickness, since the patient is being dynamically shifted with the couch during the treatment. Generally, the patient’s respiration influences the motion of tumors in the thorax or upper abdomen. During couch-tracking, the tumor motion is compensated by the motion of the robotic couch. However, the couch motion may also interact with the patient and influence the patient’s respiration, thus possibly reducing the couch-tracking effectiveness.
So far, three studies have been performed evaluating the behavior of volunteers or patients on a moving couch. Sweeney et al. [
8] conducted a study with ten healthy volunteers and 23 patients. They were positioned supine on a robotic treatment couch executing a predefined cyclical trajectory for 30 min. This procedure was repeated once on a different day. The study endpoint was the procedure termination when the patients expressed the need to stop or to administer anti-nausea agents. D’Souza et al. [
9] recruited 50 patients. They experienced several couch movement intervals that switched between static and dynamic conditions, in which the couch followed a pre-programmed trajectory. During each static segment, the ´Motion Sickness Assessment Questionnaire´ (MSAQ) [
10] was administered. Wilbert et al. [
11] performed a study with 15 healthy volunteers on a robotic couch, which counter-steered the respiration of the volunteer. The focus was on changes in volunteers’ respiration patterns.
The current study with 100 volunteers considered both the influence of respiration-driven couch motion on the respiratory pattern and the volunteers’ tolerance. The measurement alternated between static and tracking conditions in order to show whether volunteers got accustomed to tracking. In addition to the MSAQ, physiological signals (heartrate, skin conductivity, and eye motion) were recorded for an objective assessment of the mental state of the volunteers.
Discussion
Couch-tracking has the potential to reduce the treatment margins, which may lead to reduced side effects. However, one concern is the interaction between the patient and the moving couch. Here, we investigated whether the couch motion influences the respiration pattern of the volunteers and whether the couch motion induces any stress or motion sickness in the volunteers.
On average, the respiration frequency was significantly higher during tracking than during static segments. This increase is a potentially negative effect for couch-tracking, since it might cause increased compensation errors. However, during the final segment, when the couch moved along a predefined trajectory, the increase was not observed. Therefore, the increase is either due to a property of the feedback of the tracking system or of the respiration measurement. For many volunteers, the respiration measurement exhibited small oscillations above 1 Hz (for details see Additional file
1: Figure S7), which became substantial during expiration. In [
11], the significant frequency increase may not have been found, due to the small number of volunteers (15) or because their respiration measurements did not show these oscillations. The couch compensated these oscillations causing a trembling couch motion, which the volunteer might perceive as uncomfortable. Therefore, the volunteers may have tended to spend less time in the expiration. This possibility has to be investigated further, but one solution might be to use low-pass filters to mitigate these oscillations at the cost of increasing the lag time of the couch-tracking system. The flexibility of the couch and the volunteers’ bodies might lead to the occurrence of a resonant frequency, which is higher than the respiration frequency. The noise of the LTS measurement may contain frequencies in the range of the resonant frequency and together with the feedback control, it might have led to the excitation of this resonant frequency.
The respiration amplitude did not change when the couch switched between the tracking and the static segments. However, it continuously decreased over the entire experiment. This decrease may be explained by the relaxation of the volunteers. This decrease may also cause a decrease in the tumor motion amplitude. Consequently, this relaxation could be exploited by having patients rest on the couch a few minutes before starting the treatment, which results in a smaller tumor motion amplitude and in turn may lead to smaller compensation errors. In [
11], the mean amplitude over the first, the middle, and the final ten respiration cycles were computed during a five-minute measurement with the couch moving according to the respiration. For half of the volunteers, the authors found that the respiration amplitudes decreased from the first ten to the last ten cycles. This agrees with our observations. Our results additionally show that the decrease of the respiration amplitude was not affected by the couch’s switching between static and tracking conditions.
In the second part, we investigated whether the couch motion induces stress or motion sickness in the volunteers. The overall experience of the couch-tracking was evaluated using the MSAQ. The first three scores show very little evidence of motion sickness, but the fourth score (sopite related) showed higher values than the other scores. This fourth score included statements on tiredness and sleepiness, therefore, higher values of the fourth score could point to relaxation instead of motion sickness (for details see Additional file
1). The majority of the examinations took place in the evening. However, as there were outliers in all scores, a small fraction of patients might need closer observation in couch-tracking. The MSAQ was also applied in [
9], but the authors considered the overall score. Their resulting scores were generally low, which coincides with our results. Similar results were reached in [
8], where none of the subjects needed to interrupt a 30-min session of lying on a moving couch.
The skin conductivity results showed the overall relaxation (significant difference between first and last tracking segments), except for the first tracking segment, which showed a significant increase in skin conductivity emphasizing the elevated mental strain at the beginning of tracking. During the subsequent tracking segments, the skin conductivity decreased. This observation can be explained by the tracking being a new experience (first tracking segment), to which people get accustomed over the next segments. The question remains, whether people remain accustomed to couch-tracking between treatment fractions of consecutive days.
The eye-tracking showed smaller deviations from the mean gaze location during tracking segments than during static segments. The volunteers tended to focus their gaze, when the couch was moving, possibly to look for stability (analogously to looking at the horizon while balancing). However, the increase of the gaze deviations from the first tracking segment to the last tracking segment indicates that the volunteers become accustomed to tracking (for details see Additional file
1). The heartrate did not show any significant variation due to couch-tracking or overall relaxation. Since both the skin conductivity and the heartrate did not show any consistent variations with the respiration characteristics, it does not seem possible to predict the respiration characteristics using these physiological signals.
The stress of the volunteers was only slightly increased due to the couch motion (more focused gaze, increased skin conductivity). The physiological measurements agreed with the results of the MSAQ, in so far that both showed small signals of stress.
The equipment to perform the physiological measurements might have influenced the results, because, such additional devices might increase the stress level of the volunteers. However patients might have a higher level of stress, which could influence the ability to tolerate couch motion. Additionally, the gantry was static during the study, but sudden changes in the motion of the gantry during treatment could influence the stress of the patient. Patients might also have respiration patterns that are rather different from those of healthy volunteers. Additionally, the fictitious tumor motion model consisted only of a straight line. Such a model does not cover all real tumor motion trajectories.
The median age of the volunteers was 32 years, which does not reach the typical age of cancer patients, but there were 14 volunteers above the age of 60 years. However, we did not find any relationships between the aforementioned results and the volunteers’ age.
The breathing amplitudes were normalized to ensure that the fictitious tumor position stayed inside allowed motion range for the Protura. However, the normalization values were computed only on the initial static segment. Therefore, the normalization values were only approximations, which led to some variation of the resultant fictitious tumor motion amplitude (Additional file
1: Fig. S4). The LR motion was omitted due to being generally small. The LR motion could alter the patient response, since the body could rotate around the SI axis, which could be a different sensation. However, since only few patients have considerable LR motion, we neglected the LR motion.