a b s t r a c t

Objective: Some cardiopulmonary resuscitation (CPR) monitoring devices were released in recent years. Some of them are motion sensors. There are no guidelines were to position future or present sensors during CPR. We evaluate the possible influence of the location of motion sensors by a high-speed camera during a CPR on a manikin.

Material and methods: We performed a motion analysis by a high-speed camera during chest compression on a manikin to quantify chest inhomogeneous displacements and rescuer motion.

Results: Midline chest was found to have an inhomogeneous depth during CC (19 mm for the upper sternum, 27 mm for the Middle of the sternum, and 47 mm for the xiphoid). Rescuer anatomy has a complex motion.

Conclusion: The direct application of the sensor under the hand performing CC seems to be the more accurate solution if the device allows it.

(C) 2015

1. Introduction

Chest compression (CC) is one of the most important factors in cardiopulmonary resuscitation (CPR). Several parameters should be analyzed during CC: compression depth, full decompression, hand position, and frequency. According to the 2010 European Resuscitation Council Guidelines for Cardiopulmonary Resuscitation [1], an efficient CC requires 5- to 6-cm depth at a rate of 100 compressions per minute, with full chest recoil, and minimal interruptions[1]. Little is said about where this depth is actually reached.

Several audio and visual Feedback devices have been developed to assess CPR quality[2,3]. These devices measure either the pressure or movement with an accelerometer or analyze impedance changes. Some of the accelerometer-based devices use only the acceleration without retrieving the actual position. For some of the sensors, there is an algorithm to retrieve position from accelerometer data. However, the literature lacks validation studies of these devices on test benches, and there is no comparison between them, although there is a growing interest around their clinical applications [4-9]. Devices using position sensors are under development. Applications using mobile phone iner- tial unit have been proposed [10]. It is still unclear where these devices must be positioned to retrieve chest recoil. Some of these devices could be placed directly under the hand performing the CC. This position

? Conflict of interest: The authors do not have any conflict of interest.

* Corresponding author at: Intensive Care Unit, (Reanimation Polyvalente), Assistance Publique Hopitaux de Marseille, AP-HM, 264 rue Saint-Pierre, Marseille 13005 France. Tel.: +33 413499475.

E-mail address: [email protected] (S. Boussen).

ensures probably a good accuracy but could harm the chest skin or impair CC, and it is not conceivable for mobile phone devices. One could imagine putting the device on the upper chest or attaching it to the rescuer anatomy (wrist, elbow, or shoulder). But there is no guarantee that the motion recorded by a distant sensor is correlated to

CC. On other hand, the chest distortion is unlikely to be homogeneous during CC. There is not in the literature quantification of this inhomogeneity. Also, to the best of our knowledge, there is no CPR mo- tion analysis published. Thus, monitoring CPR depth is challenging: the thoracic cage motion is probably not homogeneous, and the rescuer is not fixed to the patient and has a certain degree of liberty. It is therefore important to study CC dynamics to use monitoring devices.

In this work, we performed a motion analysis by a high-speed camera during CC on a manikin. The purpose of the study is 2-fold:

-We studied the motion of different parts of manikin thoracic cage during CC to better understand distortion and relative motion of the different parts of the thorax.

-We studied the motion of the rescuer performing CC and more particularly the amplitude of the different segments of limb motion.

1. Material and methods
   1. Motion analysis and accelerometer validation during CPR

We performed a series of CCs. The manikin used was a standard manikin for CPR training (Ambu Man; Ambu A/S, Ballerup, Germany). The experimenters (4 persons: 1 woman and 3 men, all with average weight and height) performing the compressions and the manikin were equipped with miniature infrared active markers to capture the

http://dx.doi.org/10.1016/j.ajem.2015.07.051

0735-6757/(C) 2015

N. Fournier et al. / American Journal of Emergency Medicine 33 (2015) 1350-1353 1351

Fig. 1. Placement of infrared markers (black dots) during CC on the rescuer and on the manikin.

motion (Fig. 1). For the manikin, the markers were located on left eighth costal interspace, one on the left nipple, one on the left acromion, one under the xiphoid, one under the Louis angle, and the last one on the sternum’s center. Markers on the experimenter were positioned on the metacarpophalangeal joint of the lower hand, on the ulnar stiloid, one on the lateral condyle of the humerus, and the last on the greater tuberosity of humerus. Experimenters performed a total of 100 CCs (1 minute). The rescuers were asked to perform a standard CC following European Resuscitation Council guidelines, and target depth was given by the manikin mechanical indicator (green area between 5 and 6 cm). A Codamotion system (Charnwood Dynamics Ltd, Rothley, Leices- tershire, UK) was used to measure and analyze the movement by means of infrared active markers. An orthonormal coordinate system was defined by 3 fixed markers. The motion capture was performed by ODIN software (Charnwood Dynamics Ltd). Hardware data acquisi- tion was set to 100 Hz, leading to an effective data acquisition rate of

50 Hz. The measurement accuracy is 0.05 mm.

For each CPR series, we computed the average displacement ampli- tude for each Codamotion marker’s location. We computed also the angle formed by the arm and the forearm using the Codamotion system using wrist, elbow, and shoulder markers.

• 1. Data processing and statistical analysis

Data visualization and treatment were performed using Matlab soft- ware (MathWorks, Natick, MA). SigmaStat 3.5 (Systat Software/Cranes Software, Chicago, IL) was used for statistical analysis. Each measured variable value is further expressed as its mean value with its corre- sponding standard deviation interval. A linear regression looking for a correlation between the recorded motions of each part of the anatomy of the experimenter and their possible relation was performed.

1. Results
   1. Position measurements using Codamotion system

One motion-capture video performed by the Codamotion system can be found in the electronic supplementary file. The average angle between the arm and the forearm was found to be equal to 5? +- 2? for all the compression series. Fig. 2 shows the motion capture during a CPR for different locations. Displacements of the various markers are given in Table 1. Fig. 3 summarizes the displacement on the manikin and shows the different displacements along the midchest line.

Fig. 4 shows the position recorded for the sternum hand, elbow, and shoulder markers vs the wrist position. There is a linear correlation be- tween the position of the wrist and the 3 other markers. The correlation is less strong with the hand.

Table 2 gives the mean displacement of each location and the linear correlation of the displacement of the considered marker vs the wrist po- sition. The relations between the positions of the markers are not simple offsets. The case of the sternum marker is particular: the motion of the upper sternum is poorly correlated to the hand motion. The average ver- tical displacement is 19.0 +- 0.6 mm. This displacement is well below dis- placements recorded for the hand and the xyphoid displacements.

The ulnar styloid of the left hand had a similar displacement than the xyphoid (47 +- 3 vs 48 +- 7 mm, P = .08). The fifth metacarpophalangeal of the same hand was closer to the midsternum displacement (27 +- 3 vs 32+- 3 mm, P b .05).

1. Discussion

There are some new technological means to investigate the CPR depth, but motion analysis is probably more straightforward. One main issue with the use of motion sensors is where to position the

Fig. 2. High-speed camera measured the motions of different body or patient locations during 4 cycles of CPR (dashed: shoulder, x: upper sternum, plain line: hand, square: wrist, point: elbow).

1352 N. Fournier et al. / American Journal of Emergency Medicine 33 (2015) 1350-1353

Table 1

Results of the average displacement measured either by the high-speed camera or the ac- celerometer at the different sites

Location	Mean displacement (mm)	Displacement vs wrist displacement (W)	R?
Upper sternum	19 +- 0.6	0.02xW + 18	0.23
Hand	48 +- 7	0.9xW + 12.4	0.846
Wrist	61 +- 4	1xW	1
Elbow	74 +- 5	1.1xW + 7	0.93
Shoulder	83 +- 6	1.2xW + 8	0.975

* R is the correlation coefficient.

device on the body of the patient or on the rescuer or patient anatomy. For iCPR [10], it was proposed to put the smart phone on the part of ster- num that is higher than the hands performing the CPR. We showed however that the displacement of the upper sternum during CPR under- estimates the real displacement of the lower sternum and seems poorly correlated to its real displacement. Placing a device in this location will lead to underestimating the CC compression. Actually, the chest cen- tral line does not compress homogenously: we showed that when compressing the chest in between the nipple line, the chest depth in- creases from the Louis angle to the xiphoid (19 mm upper sternum, 49 mm at the xiphoid). In previous works, inspection of chest computed tomographic scan showed that the sternum is relatively fixed to the upper chest and therefore acts as a hinge [11,12]. The result is that the chest moves predominantly at the bottom end. The correlate is that CC compression is not directed toward the vertical axis but is tilted in cephal- ic direction. Thus, CC depth cannot really be defined because the displace- ment is not homogenous; moreover, devices should take into account the tilt angle and not only compute the vertical displacement. In previous work, they advocated that the upper sternum motion is constrained by the rigid support of the clavicle and the upper ribs’ particular anatomy [12]. The elastance of the upper sternum is lesser than that of the xiphoid, and compressing the sternum high needs far more rescuer energy than the xiphoid. On the other hand, caudal compression over xiphoid may lead to increased risk of internal visceral injuries.

iCPR does not calculate displacement but just gives the frequency of the

CC. It is possible to design an application that allows displacement measure- ments retrieved from the inertial measurements [13]. In this case, the upper sternum location is not suited to CC depth measurement; although it is the simplest location, depth in this location is below the actual CC. The xiphoid is probably better, but the smart phone is not in an equilibrium situation and will move. We showed that all the other parts of the chest have a differ- ent motion and the amplitude of CC is less in all locations studied in this work.

Fig. 4. Displacement of the lower sternum, hand, elbow, and shoulder infrared markers vs the wrist displacement during CPR as recorded by a high-speed camera.

Another solution is to position inertial measurement devices by fix- ing it to the wrist, the elbow, or even the shoulder of the rescuer. We have shown in this study that the movements of the various member segments during CPR are different even when CPR is performed by an experienced professional and when the angle between the arm and the forearm is locked to 5?. The farther of the sternum the marker is, the greater the measured range of motion. This difference should be considered if one wants to attach the device on the rescuer arm. It would require a corrective factor depending on the position.

The more accurate position seems to be the hand or directly on the lower part of the sternum. The measure will be less prone to errors if the sensor is placed closer to the sternum. However, the hand motion is complex, and the caudal contact point will have larger amplitude than the cephalic contact point.

The center of the chest has no homogeneous motion, and therefore its displacement is not easy to quantify. Literature indicates that 5 to 6 cm should be reached, but it does not say precisely where. The nipple line seems to be a good location.

The 3-dimensional motion reconstitution of movement used in this may also allow rescuers to visualize their movement during CPR to im- prove their practice. However, this technology is too complex to have any clinical application except for research or educational purposes unless

Fig. 3. Summary of the displacement of different parts of the chest during CC. The depth of the compression increases with the caudal displacement. The chest distortion is therefore highly inhomogeneous.

N. Fournier et al. / American Journal of Emergency Medicine 33 (2015) 1350-1353 1353

Table 2

Mean displacement of different chest locations during CC

Location Displacement (mm)

Acromyon 0.6 +- 0.5

Nipple 12.8 +- 0.5

Left 6th intercostal space 23.2 +- 1

Upper sternum 19.0 +- 0.6

Middle sternum 27 +- 3

Xiphoid 47 +- 3

miniaturization makes it possible to use it at the field, and motion detec- tors as accelerometers are expected to be more convenient during CC.

1. Conclusion

During CC, the chest motion is not uniform by far. Along the sternal line, motion depth is very different. Our study shows that a CC measuring device could not be placed anywhere, as the amplitude of the CC motion is dependent on the considered site. We show in this work that the chest motion is not homogenous and the upper sternum moves significantly less than the xiphoid. Placing a device on the rescuer leads to overestima- tion if placed far from the chest. On the other hand, the rescuer motion during CC is not constant through the rescuer anatomy. Therefore, the direct application of the sensor under the hand performing CC seems to be the more accurate solution if the device allows it.

Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.ajem.2015.07.051.

References

1. Nolan JP, Soar J, Zideman DA, Biarent D, Bossaert LL, Deakin C, et al. European Resus- citation Council guidelines for resuscitation 2010 section 1. Executive summary. Re- suscitation 2010;81:1219-76. http://dx.doi.org/10.1016/j.resuscitation.2010.08.021.
2. Lukas RP, Van Aken H, Engel P, Bohn A. Real-time feedback systems for improve- ment of Resuscitation quality. Anaesthesist 2011;60:653-60. http://dx.doi.org/10. 1007/s00101-011-1909-9.
3. Semeraro F, Frisoli A, Loconsole C, Banno F, Tammaro G, Imbriaco G, et al. Motion detection technology as a tool for cardiopulmonary resuscitation (CPR) quality training: a randomised crossover mannequin pilot study. Resuscitation 2013;84:501-7. http://dx.doi.org/10.1016/j.resuscitation.2012.

12.006.

1. Zapletal B, Greif R, Stumpf D, Nierscher FJ, Frantal S, Haugk M, et al. Comparing three CPR Feedback devices and standard BLS in a single rescuer scenario: a randomised simulation study. Resuscitation 2013. http://dx.doi.org/10.1016/j.resuscitation. 2013.10.028.
2. Yeung J, Meeks R, Edelson D, Gao F, Soar J, Perkins GD. The use of CPR feedback/ prompt devices during training and CPR performance: a systematic review. Re- suscitation 2009;80:743-51. http://dx.doi.org/10.1016/j.resuscitation.2009.

04.012.

1. van Berkom PFJ, Noordergraaf GJ, Scheffer GJ, Noordergraaf A. Does use of the CPREzy involve more work than CPR without feedback? Resuscitation 2008;78: 66-70. http://dx.doi.org/10.1016/j.resuscitation.2008.01.024.
2. Thomas SH, Stone CK, Austin PE, March JA, Brinkley S. Utilization of a pressure- sensing monitor to improve in-flight chest compressions. Am J Emerg Med 1995; 13:155-7. http://dx.doi.org/10.1016/0735-6757(95)90083-7.
3. Skorning M, Beckers SK, Brokmann JC, Rortgen D, Bergrath S, Veiser T, et al. New vi- sual feedback device improves performance of chest compressions by professionals in simulated cardiac arrest. Resuscitation 2010;81:53-8. http://dx.doi.org/10.1016/j. resuscitation.2009.10.005.
4. Pozner CN, Almozlino A, Elmer J, Poole S, McNamara D, Barash D. Cardiopulmonary resuscitation feedback improves the quality of chest compression provided by hos- pital health care professionals. Am J Emerg Med 2011;29:618-25. http://dx.doi.org/ 10.1016/j.ajem.2010.01.008.
5. Semeraro F, Taggi F, Tammaro G, Imbriaco G, Marchetti L, Cerchiari EL. iCPR: a new application of high-quality cardiopulmonary resuscitation training. Resuscitation 2011;82:436-41. http://dx.doi.org/10.1016/j.resuscitation.2010.11.023.
6. Pickard A, Darby M, Soar J. Radiological assessment of the adult chest: implications for chest compressions. Resuscitation 2006;71:387-90. http://dx.doi.org/10.1016/j. resuscitation.2006.04.012.
7. Shin J, Rhee JE, Kim K. Is the inter-nipple line the correct hand position for effective chest compression in adult cardiopulmonary resuscitation? Resuscitation 2007;75: 305-10. http://dx.doi.org/10.1016/j.resuscitation.2007.05.003.
8. Oh J, Song Y, Kang B, Kang H, Lim T, Suh Y, et al. The use of dual accelerometers im- proves measurement of chest compression depth. Resuscitation 2012;83:500-4. http://dx.doi.org/10.1016/j.resuscitation.2011.09.028.