Evaluation of natural head position over five minutes: A comparison between an instantaneous and a five‐minute analysis with an inertial measurement unit

Abstract Background Head posture is a balance of several positions and therefore shows inherent variation. Most methods available to quantify this are however instantaneous, not providing information about its variation over time. A dynamic recording of head posture would thus be beneficial. Objectives The purpose of this study was to evaluate the variation in natural head position (NHP) over 5 min using an inertial measurement unit (IMU). Methods Fifteen healthy young volunteers were asked to sit on a chair and keep their head in the self‐balanced position for 5 min. A mirror was then revealed in front of them, and they were asked to look at their eyes for 20 s. This procedure was undertaken on two separate occasions with a one‐week interval. This was compared to an instantaneous measurement of head position at a specific time point corresponding to the 15th second of the recording. Results During the 5 min of recording, the participants tended to elevate their head progressively by a mean of 1.5°, which is then corrected by looking at oneself in the mirror. Most participants tended to rotate their head to the left and continued that progressive rotation despite looking in the mirror. The roll axis had no systematic changes observed between the self‐balanced position and the mirror‐guided position and was the most reproducible axis. Moderate to good correlations were found comparing both sessions for each axis. Conclusion The comparison between the five‐minute analysis and the instantaneous measurement showed a systematic difference on the pitch axis but no differences for the yaw and roll. These results suggest that the variation in the NHP during a period of 5 min is generally specific to each participant with a head elevation and rotation to the left in most cases.


| INTRODUC TI ON
Most methods available to measure the natural head position (NHP) 1 use an instantaneous measurement corresponding to one position at a specific point in time. However, the NHP is known to be a balance of several positions 2 ; therefore, it is not a fixed position, and inherent variation is present for each person. The characteristics and implications of this variation are not fully understood.
The head position is related to different alterations in fields such as orthodontics, 3 orthopaedics, 4 respiratory physiology, 5,6 and ophthalmology. 7 To measure the NHP, there are two main standard positions, namely, the self-balanced position and the self-guided position. The self-balanced position is adopted when a person looks at an imaginary point in front of him/her, from an infinite distance at the eye level. 8 The self-guided position is when a person looks at themselves in a mirror. 8 Different studies indicate that the mirror-guided head position is the most stable and reproducible 9,10 ; however, it is rarely adopted in everyday life. It is thus preferable, in order to more accurately simulate the head position in a natural environment, to analyse the self-balanced head position that is more frequently adopted.
In growing individuals, there are hypotheses that the head position may influence the craniofacial growth pattern. 8,11 In such cases, it is possible that the evaluation of the NHP during a longer period could be of greater interest rather than solely looking at the head position at a specific point in time.
Evaluation of the NHP is performed on three axes namely pitch, roll, and yaw. The pitch is the angle between the horizontal plane and the Frankfort plane. The roll is the angle between the horizontal plane and the bipupillary plane. Finally, the yaw is the angle between the acromial process and the median sagittal plane. (Figure 1).
The aims of this study were to 1. Compare the instantaneous analysis to a five-minute analysis of the head position.
2. Analyse the evolution of the NHP in a self-balanced position for 5 min and compare it to the mirror-guided head position for the three axes.
3. Evaluate the reproducibility of the self-balanced head posture and the mirror-guided head position for the three axes in a group of healthy young adults.

| Ethics statement
The present study was approved by the Swiss Association of Research Ethics Committee. The experimental procedures were conducted in conformity with the Declaration of Helsinki. Informed consent for participation in the study and publication in an openaccess format was obtained from the participants, with regard to their recordings and personal information. The procedures of the study were fully explained to the participants, and they provided their informed written consent prior to being enrolled in the study.

| Subjects and methods
Fifteen healthy young adult volunteers (eight women and seven men) at the University Clinics of Dental Medicine in Geneva, Switzerland, aged from 20-30 years (mean age of 28.4 years) were recruited for this study.
Wearable sensors or inertial measurement units (IMU)s were used to measure head position. These are electronic devices allowing movements to be tracked in three dimensions (3D). 12 This system was chosen because it is easy to use, precise, 13 and allows tracking of head position in a given period. This system is composed of a gyroscope, an accelerometer, and a magnetic angular rate and gravity. The hardware consists generally of two parts, a detector fixed on the element of interest and a receiver. The information is stored to allow one to access the recordings of the patients and to follow them. 14 An IMU allows the head position to be recorded dynamically for a prolonged time period, thus it is possible to analyse the full variation of the NHP during a 5-minute recording.
The participants were asked to sit on a chair in the 90-90-90 position (corresponding to a 90° angle on the hips, the knees, and the feet) 15 facing a white wall which was one meter in front of them.
They were asked to look in front of them and to keep a straight head posture with their own feeling of natural head balance (self-balanced position). This position was recorded for 5 min while listening to a short story played behind them with stereo-speakers carefully centred on the volunteers. Following this, a mirror was placed on the wall in front of the participants, and they were asked to look at their eyes for 20 s (self-guided position). This procedure was repeated twice with a one-week interval, and during the second recording, a different short story was played in order to allow for a similar experimental setup without the participants having to listen to the same story twice.
This system is composed of two sensors. Before measuring the head position, this system needs to be calibrated to assess the neutral position (corresponding to 0° on each axis). This was done by placing the first sensor on the ground in front of the participant and the second one behind them. Both have to be parallel, on the same plane, and perpendicular to the ground axis. After calibration, the first sensor was placed on the forehead of the participant.
In order to minimize the error due to the positioning of the sensor, one front face photograph was taken to make sure that the sensor was parallel to the bipupillary line. A second profile photograph was taken to eliminate the slope of the forehead from the IMU recordings. The postprocessing calibration was performed with Pixelstick (version 2.16.2, Plum Amazing software LLC). This calibration method was the same as that used by Billiaert et al. 16 that compared the IMU system to lateral photographs and showed excellent accuracy.

| Statistical analysis
All statistical analyses were performed using SPSS (version 24.0, SPSS Inc.). The data were separated into four different segments.
The initial segment (I) represents the mean of the first 20 s, the finishing segment (F) the mean of the last 20 s before the mirror-

| RE SULTS
The data show a greater interindividual variation in the pitch axis with a negative value corresponding to an elevated Frankfort plane than the horizontal plane. The roll axis had the least amount of interindividual variation and a slight deviation to the right compared to the median plane. The yaw axis had comparable interindividual variation to the roll axis and had a slight deviation to the left compared to the plane that passes by the acromial process (shoulders) ( Table 1).

| Comparison of the one-shot procedure to the five-minute analysis
When comparing the instantaneous analysis to the mean of the fiveminute analysis, a systematic difference was found for the pitch axis (p = .04), whereby the mean pitch from the five-minute procedure was elevated by 0.9 degrees compared to the instantaneous procedure. For the roll and yaw axes, no systematic differences (p > .05) were found. However, a good correlation and good to excellent reliability was found for all the three axes ( Table 2). Overall, great correlation was found between the different segments for the pitch and roll axes ranging from 0.72 to 0.87. However, for the yaw axis, the correlations were lower, ranging from 0.49 to 0.93 due to higher deviations. For the pitch and roll axes, good-toexcellent reliability was found between the segments (ranging from 0.83 to 0.93), and for the yaw axis moderate to excellent reliability was found (ranging from 0.55 to 0.96).

| Evaluation of the reproducibility of the selfbalanced and the mirror-guided head position
No systematic differences were found between the two sessions for each segment (p > .05). When comparing the same segments for the two different sessions (Figure 3), moderate to good correlations were found depending on the axis. The pitch axis had the strongest correlations followed by the roll and the least reproducible axis was the yaw. Good to excellent reliability was found for the self-balanced head position (ranging from 0.78 to 0.96) and poor-to-excellent reliability for the mirror-guided head position (ranging from 0.49 to 0.92) depending on the axis ( Table 3).

| DISCUSS ION
The present study showed an excellent correlation between the instantaneous procedure and the mean of the five-minute recordings for the pitch and roll axes and good correlation for the yaw axis. The five-minute analysis showed a systematic elevation of the head compared to the instantaneous procedure, but no systematic changes for the roll and yaw axes were observed.
The data recorded at the 15th second with the IMU system were used as a representative instantaneous procedure. This time was chosen because it represents approximately the time required to take a photograph. Moreover, unlike photography, the IMU system allows one to obtain data for all three axes. The study of Billiaert et al. 16 as well as Al-Yassary et al. 19 demonstrated excellent accuracy of the IMU system for measuring head position.  Note: Shown are the mean and standard deviation (SD) of the difference (T 1 -T 2 ) between the instantaneous (corresponding to T 1 ) and the five-minute recording (corresponding to T 2 ), Pearson's correlation (r), the intraclass correlation coefficient (ICC) 17 with 95% confidence interval (95% CI), and their standard error measurement (SEM) for each axis.

TA B L E 2 Comparison between an
instantaneous and a five-minute recording of head position using an inertial measurement unit When the different segments were compared, it was found that the participants tended to elevate their heads and turn progressively to the left. However, a general trend for the roll axis was not found. When the mirror was displayed, it was observed that the participants corrected their elevation on the pitch axis, without any effects on the other two axes (roll and yaw). The participants that tended to rotate their head kept their rotation and continued in the same direction.
The variation of head position seems to be influenced by differ- For the mirror-guided head position, a good correlation was observed on the pitch axis since the mirror affects it. However, the mirror-guided position does not seem to influence the roll and yaw axes, which is why the correlation was much lower for these axes.
For the self-balanced head position, good correlation was found even for the roll and the yaw axes. This indicates a similar pattern between the two sessions, meaning that the variation in head position may be specific to each participant.
The limitations in the present study include the potential lack of generalizability as all the included participants were healthy young adults and, thus, represent only a small part of the population.
However, the use of this system can be interesting for every population including children, the elderly, and patients with a handicap or other pathological conditions since it is easy to use and allows a recording of the variations of head position over a certain period of time.
The reproducibility found in this study is based on a one-week interval, and on two sessions. A longer period between the recordings and multiple sessions would be necessary to evaluate the longterm reproducibility. Finally, different elements influence the head position, such as fatigue and ocular and respiratory factors, and thus the recordings of each participant should be evaluated when these factors are the same for the two sessions.

| CON CLUS ION
Good-to-excellent correlations were found between the instantaneous analysis of head position and the five-minute head position recordings. However, on the pitch axis, the head position was systematically more elevated in the five-minute analysis than the instantaneous procedure, whereas no difference was observed for the roll and yaw axes.
Most participants have the same pattern of evolution of their head position over a five-minute recording. An elevation of the head and a rotation to the left were observed as general patterns. Only the elevation of the head, however, is corrected by the mirror-guided position.
The pattern of evolution of the head position and its variation are generally specific to each participant with moderate-to-good correlation between the two sessions. Overall, the means of the five-minute recordings were more reproducible than those of the mirror-guided head position. The roll axis had the least amount of variation, followed by the pitch and finally the yaw axes, respectively.

ACK N OWLED G EM ENTS
The authors received no specific funding for this work. Open Access Funding provided by Universite de Geneve.

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

AUTH O R CO NTR I B UTI O N S
K.B. and M.A., contributed to the data collection and data analysis, G.S.A. and S.K. contributed to the study design and drafting of the manuscript.

PE E R R E V I E W
The peer review history for this article is available at https://publo ns.com/publo n/10.1111/joor.13297.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data included in this study are available upon reasonable request from the corresponding author.