Comparison of different optical techniques to measure skin vibrations

We test different possible setups to measure skin vibrations in the abdominal region which could, in the future, enable quick and comfortable monitoring of aortic aneurysms in the abdominal region. For this study, a simple mechanical model is used that simulates the human tissue and a large blood vessel, the aorta. The model consists of a plastic box filled with a gelatin as used for ballistic experiments. A rubber hose goes straight through the gel block and, with a pump that is connected to the hose, one can create a pulsatile flow of water through the flexible hose leading to traveling waves along the artificial vessel. Surface vibrations on the gelatin block are then measured using a Laser-Doppler vibrometer (LDV), a commercial 3D DIC system and a simple camera setup with only one camera and an additional mirror for 3D capabilities. While the LDV offers the best signal quality, it can only measure one point simultaneously. We discuss implications how this will influence the possible data analyses compared to more noisy, but full-field measurements provided by the camera-based systems.


Introduction
Abdominal aortic aneurysms (AAA) are a severe medical condition that primarily affects men and women above 60, with men being significantly more likely to have an AAA than women [1,2].As the blood vessel tissue loses its elasticity and strength as people age, the aorta walls may bulge locally under systemic blood pressure.Most often, this happens in the abdominal region.Such an AAA may not cause any symptoms initially, but as it grows, the risk of rupturing increases, and patients can experience unspecific pain.Furthermore, a potential rupture is a severe threat, so doctors recommend a repair when the aneurysm reaches a size of around 5 cm or grows at an alarmingly high rate.A complete replacement of the aneurysm using a synthetic tube sewn in place is the traditional way to repair an AAA.However, as this technique requires severe surgery during which the patient's abdominal region must be fully open, a minimally invasive alternative has been developed and has gained more and more share in total aneurysm repairs over the past years.This technique is called endovascular aneurysm repair (EVAR) and is carried out using a special stent with an additional tissue layer.This so-called stent graft can be deployed via the leg arteries, so patients can typically leave the hospital two days after the procedure.The stent graft sits inside the aorta and acts as a tunnel through which the blood flows, and it is held only by friction between the graft and the aorta wall, sometimes also with the help of small hooks at the outside.This mechanism is why regular check-ups after EVAR are mandatory for patients, as leakages between the stent graft and the aorta wall can occur.Also, through smaller blood vessels, blood can reach the aneurysm sac.No matter the reason, an aneurysm exposed to systemic blood pressure is at risk of growing further, so patients undergo regular checks post-EVAR.To this end, doctors use imaging techniques to observe the current state of an aneurysm, such as computer tomography (CT) or contrast-enhanced ultrasound (CEUS).Both options involve either harmful procedures (X-rays during a CT scan) or uncomfortable (injection of contrast media for CEUS) for the patient.
At this point, our approach comes into play.We want to investigate the feasibility of contactless measurements of patients' abdominal skin to identify aneurysms and indicate whether they got pressurized after EVAR to reduce the need for more invasive state-of-the-art methods.
in the past, preliminary tests to measure the abdominal skin motion with a laser Doppler vibrometer and a 3D DIC system were conducted with some success [3,4].Here, we propose a third method using a single high-speed camera and we qualitatively compare the measurements.

Experimental Setup
This section will explain the experimental setup, which involves the test subject imitating the human abdominal region with blood flow through it, and the different measurement devices that will be evaluated in this study.

Gelatin Block
The main test subject consists of an artificial "blood circuit" that is, in part, suspended in a block of highly concentrated gelatin (fig.1).The blood circuit comprises a rubber hose (inner diameter 8 mm, wall thickness 2 mm) and different pieces of silicone tubes that are connected to the tube via plastic pieces.With a small hand pump attached to the hose, a pulsatile flow of water can be induced by pressing the pump.The actuation of the pump has been automated via a motor-driven mechanism that squeezes the pump by pushing a cylinder through an eccentric disc attached to the motor shaft.The free ends of the rubber hose reach into a water reservoir (a large drinking glass) such that the fluctuating volume of water in the pump-hose system can be compensated for.
The flat surface of the gelatin model is where we will measure the out-of-plane motions along the hose.Even though neither the consistency of the gelatin and water nor the elasticity of the silicone tubes will match precisely the properties of human tissue and blood, it is expected that this setup can be capable of exhibiting the same basic phenomena as the human skin when exposed to a pressure-induced flow of fluid through the model.

Measurement Equipment
After explaining the test subject, we shall look at the measurement devices that are compared in this study.All of them have in common that the measurement can take place without direct contact with the test specimen, which is desirable when we think about the possible future application in patients where this means a higher level of comfort during the examination.
Figure 2 illustrates the measurement setups that are later compared.This view from the side shows how we usue a mirror to redirect light for two of the measurement strategies, which is explained in more detail in the following.

Scanning Laser-Doppler Vibrometry
A laser Doppler vibrometer (LDV) is a device that can measure motion along the direction of a laser beam on surfaces that reflect this laser light.In this study, we use a Polytec RSV-150 vibrometer, which uses a 1550 nm wavelength laser beam.
Figure 1: Gelatin block with the two artificial vessels inside.Currently connected is the healthy aorta, while the aneurysm is inactive.Reflective stickers (LDV) and stickers with a black and white speckle pattern (cameras) are respectively used to enable the measurements along the vessels.The perspective is quite representative of the high-speed camera's view and creates two different viewing angles onto the points of interest -one almost perpendicular view through the mirror and one direct view at an angle.Simultaneously, the mirror is used to redirect the laser beam of the LDV to be able to mount it as usual on a tripod and still measure the out-of-plane motion of the gelatin.The 3D DIC system records the measurement points directly through its two cameras without using the mirror.Additionally, to measure multiple points along the gelatin block in a reasonable amount of time, the LDV was equipped with a scanning unit (Maul-Theet Scan-Set).This device redirects the laser beam using two mirrors, each of them rotating about one axis such that the measuring spot can be moved in the region of interest.Through the accompanying software, a measurement grid can be defined on the subject's surface, which is then, point by point, scanned by the device to gather all desired measurements.Such an arrangement is usually called a scanning LDV (SLDV).
Here, an additional mirror (fig.2a) is used to redirect the measurement beam such that a usual horizontal mounting of the SLDV can be used to measure the vertical motion of the gelatin surface.

AIVELA-2023
Journal of Physics: Conference Series 2698 (2024) 012021 IOP Publishing doi:10.1088/1742-6596/2698/1/0120214 2.2.2.3D DIC Secondly, we measure the surface vibrations using a commercial 3D DIC system (GOM ARAMIS SRX with a 300 mm profile) as illustrated in Figure 2b.Its two cameras are located side-by-side at an angle such that the measured points are ideally located at a distance of 70 cm to the center of the camera profile.For the measurements with the DIC to work correctly, square stickers of 10 mm × 10 mm with a printed speckle pattern were used, which allowed the software to identify the respective points in both cameras in all captured frames.

High-Speed Camera Setup with Mirror
The third means of measurement is a high-speed camera (HSC) used in conjunction with a mirror through which a second view of the points of interest is projected on the camera sensor, see fig.2c.This setup enables us to capture threedimensional information with a single camera.To this end, we first identify 2D displacements in the image sensor's plane using the Lucas-Kanade method ( [5]) through the open-source toolbox pyIDI 1 .In the second step, we combine the displacement information of the measurement points from both the direct and the mirrored views.Here, a simplifying assumption is used: expressly, we assume that the camera is relatively far away from the object using a long focal length lens.Under these circumstances, it becomes acceptable to consider that all the incident light reaches the sensor in parallel.Then, the geometric parameters that define how to interpret the measured displacements in the sensor plane are reduced to only three rotation angles and one conversion factor (px → mm) per view.One can use one frame to find these parameters during the measurements in which the world coordinate axes can be seen.For details concerning this reconstruction, refer to [6] Here, we will only consider the out-of-plane motion of the gelatin surface for the comparison.For future studies, however, also in-plane motion could be of interest.

Conducting the Measurements
We sequentially conducted measurements with the artificial aneurysm and with the healthy aorta replica to compare the three measurement techniques.
For each measurement run, we activated the mechanism that runs the pump and waited for a few seconds until the motor was running at nominal speed and the flow of water seemed regular.Then, the respective measurement device started recording for several seconds such that at least five pulses were captured in each run.The camera-based systems ran at 1000 fps while we chose a sampling frequency of 1280 Hz for the SLDV as the same rate as for the cameras is not available in the software.

Results
Now, we use the described experimental setup to compare the three measurement techniques under consideration and investigate differences in the signal when using the aneurysm or the healthy aorta.As mentioned, only one acquisition device (LSDV, 3D DIC, high-speed camera) was used simultaneously, so the actual displacements are different.However, using the automatic pump actuator creates highly reproducible results such that we will synthesize a typical beat pattern by averaging multiple pulses in time for each measurement run that comprises multiple "heart" beats.
First, we will look at how different the measurements look, specifically, how much the camerabased techniques suffer from a higher noise floor than the laser vibrometer.
Figure 3a showcases the vibration pattern of the midpoint when the aneurysm is connected to the water circuit.There is practically no observable difference in the motion patterns captured by the different devices.Slight differences in the base frequency are the result of the three lines not coming from the exact same measurement, but being captured in three different runs.
As the driving motor is not controlled in speed, but the power supply was simply switched on without manipulating the voltage between the runs, it is not guaranteed that the exact same rotor speed is achieved.One possible reason for slight changes in the rotor speed could be the relatively high friction in the pumping mechanism which is likely subject to some fluctuations as the 3D-printed material wears off.
Other than that, amplitudes are well captured, and the signal noise is well contained for the camera-based techniques.The latter is better visible in a more detailed plot, given in fig.3c.Here, we see that finer details, like the slight hump in the falling slope of the second peak, are nicely captured within all measurements, which also gives confidence in the repeatability of the measurements.A similar impression is gained when we look at the corresponding displacements  in the regular cylindrical aorta (figs.3b and 3d).Here, the amplitudes are much more variable from pulse to pulse, which is why the measurement techniques are also more complicated to compare.Still, no fundamental problem is recognizable with any of the three techniques, and the three-peak displacement pattern is well seen in each plot.While LDV and the commercial 3D DIC system are proven techniques, we should look at the results of the simplified 3D reconstructed signal gained from the high-speed camera more closely.To this end, we identify the first peak of each pulse pattern and align multiple pulses at this point to overlay these time signals.Figure 4 displays this overlay for both the aorta and the aneurysm, and on top, the dark blue line represents the mean curve of the eight separate pulses.The excellent repeatability that could already be assumed after looking at fig. 3 is now very obvious, especially for the aneurysm (fig.4a).But also for the regular aorta (fig.4b), the signals share the same characteristics, and there is no sign of major issues regarding random errors of this measurement technique other than a noise floor which, however, doesn't affect the overall shape of the pulses.

Discussion
The results presented in the previous section strongly indicate that any of the contact-less measurement strategies under investigation can capture motion induced by a fluid flow through a flexible, artificial blood vessel.
Finally, we should take a very brief look at something that needs more attention in future studies: the time delay between the signals along the silicone vessels.This time delay should, in theory, be the result of the speed of the pressure waves traveling along the tubes, a value called pulse wave velocity (PWV), which is already used as an indicator for the aging of the vascular system [7] and could also be relevant for the monitoring AAAs after repair as a defect might affect its value around the aneurysm.
To this end, we looked for apparent time delays, which proved more difficult than hoped.Due to the short distance from the first to the last measurement point, the delay is expected to be in the range of some milliseconds.The comparatively low-frequency motion of the surface makes it hard to reliably identify such a short delay.
Figure 5 shows one pulse captured by the 3D DIC system.While most of the pulse's shape is dominated by quite low frequency, there is one detail that is localized enough in time to determine the delay: right after the second main hump in the signal's profile, a small peak follows.The intermediate local minimum is visible in all five curves representing points along the aneurysm.Also, it does not seem to be caused by signal noise, as it is equally visible in all the other pulses of the measurement run.Based on this observation, a PWV of a little over  5 m s −1 can be read from the plot for the aneurysm.This value would be in the same region as the real PWV in a human aorta [7].
Unfortunately, the pulse shape for the healthy aorta doesn't show such a repeated fine detail, which is why we cannot give a reliable value to compare the aneurysm and aorta in this respect.

Conclusion
We have shown that, in principle, all three methods under investigation -3D DIC, SLDV, and the own in-house camera/mirror setup -can be applied to future measurements.The observed accuracy of the camera/mirror combination is encouraging, as this represents the potentially cheapest version.Of course, the quality of an LDV measurement in terms of signal-to-noise ratio will never be matched by either of the camera-based systems.However, it is limited to only one measurement point at a time, bringing many complications.This is especially true when we will observe the time histories of single points and the spatial characteristics of the traveling waves along the abdominal region.Very precise triggering will be necessary to synchronize the single measurements and reveal the expected slight delays caused by the pressure wave traveling through the elastic vessel.Finally, for a future application of monitoring AAAs after repair, the amplitudes should be at least similar to what we saw on the gelatin block in the present study.A first measurement on a single test person's abdominal region showed vibrations with roughly the same amplitude as we saw for the regular aorta, i. e. around 0.2 mm.
With these positive results on the in-vitro model, the next step will be to conduct a similar study on several healthy test subjects to gather data and investigate whether similarly repeatable motion patterns can be observed.Furthermore, we need to take a more detailed look at the time delay between the measurement points along the tubes to capture traveling wave characteristics, which could also contain important information about the state of a repaired aneurysm as the pulse wave velocity is expected to be a possible indicator for complications after surgery.

Figure 2 :
Figure 2: Schematic side view of the different measurement setups.In this view, the box with gelatin is cut such that the vessels transport the flow of water perpendicularly to the image plane.

Figure 3 :
Figure 3: Comparison motion patterns as measured by SLDV, 3D DIC, and high-speed camera setup for the respective midpoint of the artificial blood vessel.

Figure 4 :
Figure 4: overlay of eight consecutive pulses () as measured by high-speed camera, aligned at their first peaks, and their mean curve ().
Local minima after second main peaks

Figure 5 :
Figure 5: One pulse and a small detail towards the end for the aneurysm measured by the 3D DIC system.Multiple points along the vessel are plotted with lighter colors further downstream.