In vivo measurement of the biomechanical properties of human skin with motion-corrected Brillouin microscopy

Biomechanical testing of human skin in vivo is important to study the aging process and pathological conditions such as skin cancer. Brillouin microscopy allows the all-optical, non-contact visualization of the mechanical properties of cells and tissues over space. Here, we use the combination of Brillouin microscopy and optical coherence tomography for motion-corrected, depth-resolved biomechanical testing of human skin in vivo. We obtained two peaks in the Brillouin spectra for the epidermis, the first at 7 GHz and the second near 9-10 GHz. The experimentally measured Brillouin frequency shift of the dermis is lower compared to the epidermis and is 6.8 GHz, indicating the lower stiffness of the dermis.


Introduction
Skin aging and various diseases such as psoriasis, scleroderma and erysipelas affect skin mechanics [1].Therefore, studies of skin elasticity are potentially useful in assessing tissue disease and the efficacy of medicaments and cosmetic products in terms of emollience and hydration.The mechanical properties of the skin are determined by the thickness and quality of the epidermis, dermis and subcutis (see Fig. 1).The epidermis, especially the stratum corneum, is a tight barrier against the external environment.It is tough and resilient due to the presence of fibrous keratin [2,3].The dermis layer is 1 mm thick and consists of a network of collagen with interspersed elastic fibers and lymphatic elements [2].

Introduction
Skin aging and various diseases such as psoriasis, scleroderma and erysipelas affect skin mechanics [1].Therefore, studies of skin elasticity are potentially useful in assessing tissue disease and the efficacy of medicaments and cosmetic products in terms of emollience and hydration.The mechanical properties of the skin are determined by the thickness and quality of the epidermis, dermis and subcutis (see Fig. 1).The epidermis, especially the stratum corneum, is a tight barrier against the external environment.It is tough and resilient due to the presence of fibrous keratin [2,3].The dermis layer is 1 mm thick and consists of a network of collagen with interspersed elastic fibers and lymphatic elements [2].Several methods based on mechanical models have been used to measure the elastic properties of human skin in vivo.The common approach is to combine a mechanical system, such as a Several methods based on mechanical models have been used to measure the elastic properties of human skin in vivo.The common approach is to combine a mechanical system, such as a suction, indentation, torsion or twisting system, with an ultrasound scanner for simultaneous visualization of the deformation of the skin structure in vivo [4][5][6].Ultrasound elastography based on shear wave dispersion measurements has been used for skin elasticity testing [7][8][9].However, it is not possible to resolve the thin skin layers such as the epidermis using mechanical or ultrasoundbased techniques due to the limited spatial resolution.In addition, it is important to measure skin thickness simultaneously to obtain correct elasticity values, since the mechanical response of the skin depends not only on Young's modulus but also on skin thickness.Optical coherence elastography (OCE) [10][11][12][13][14][15][16] provides high-resolution imaging down to a few micrometers and allows assessment of skin elastic properties, but also requires mechanical stress to be applied to the skin.To perform a non-contact, all-optical test of skin elasticity, the Brillouin imaging approach has been used [17,18].Brillouin spectroscopy is a label-free and non-contact technique that can assess the viscoelastic properties of biological samples via photon-phonon scattering interactions [19][20][21].Using Brillouin microscopy in combination with refractive index measurements [22], it is possible to obtain the longitudinal elastic modulus locally with micron resolution and without any external mechanical deformation (see Fig. 1).This technique is promising for testing the elastic properties of skin layers.
Brillouin microscopy has previously been used for ex vivo skin biomechanical testing [17,18] using a light source with a wavelength of 532 nm, which cannot penetrate deep into the skin and reach the dermis.The use of longer wavelengths may be promising for biomechanical testing of both the epidermal and dermal layers and for obtaining a depth profile of skin modulus.For in vivo measurements, it is important to follow the movement of the skin during the procedure, as sample movement cannot be avoided.The combination of Brillouin microscopy and optical coherence tomography (OCT) allows elasticity to be measured with the necessary structural guidance [23,24], eliminating the influence of sample motion [25,26].
In this paper, we performed the depth-resolved motion-corrected mechanical mapping of human skin in vivo using the combination of Brillouin microscopy and OCT.By tracking the sample position using OCT, we were able to detect differences in the Brillouin spectra of the epidermis and dermis.Interestingly, we observed two peaks in the Brillouin spectra of the epidermis: at ∼7 and ∼10 GHz.We assume that the higher peak corresponds to the keratinized cells in the stratum corneum.In the dermis, the Brillouin frequency shift is lower, around 6.8 GHz.

Method
A schematic of the optical setup is shown in Fig. 2(A).It consists of two parts: the Brillouin imaging system and the OCT-based depth tracking system.

Brillouin imaging system
The linearly polarized light from the CW laser (Cobolt Flamenco, 660 nm) is coupled to a polarization-maintaining fiber and then collimated to a 6.7 mm diameter beam using a collimator (Thorlabs, F810FC-635).The laser beam passes through the polarizing beam splitter (PBS) and the quarter-wave plate (QWP) and is focused by an 8 mm focal length lens with an NA of 0.5 (Thorlabs, A240TM-B).The laser power before the imaging lens is less than 20 mW.After scattering in the sample, the light is collected back and reflected by a PBS, coupled to the fiber by a collimator (Thorlabs, F810FC-635), and directed to the VIPA-based spectrometer (HF-8999-660-AUTO, Light Machinery).In the spectrometer, the Rayleigh scattering signal is reduced by a tunable etalon filter with 55 dB suppression, and the Brillouin spectra are obtained using a single VIPA and 2D camera.Typical spectra of light scattered by the sample are shown in Fig. 2(C).

OCT-based depth tracking system
For OCT-based depth tracking, light from a superluminescent diode (SLD-371, Superlum, Carrigtohill, Ireland) with a center wavelength of 840 nm and a full width at half maximum bandwidth of 52 nm was used.The light from the SLD was coupled to a fiber-based beam splitter (BS, Thorlabs, TW850R3A2) and split into sample and reference arms.In the sample arm, the light was collimated to a 2 mm diameter beam and directed to the imaging lens by a dichroic mirror (DM, Thorlabs, DMSP750B).The light scattered by the sample was collected back and split with the reference beam.The interferogram was recorded by the home-built spectrometer.
Standard OCT processing was used to obtain the depth dependence of tissue reflectivity (A-scan).
The signal-to-noise ratio (SNR) for the system is approximately 90 dB and the SNR roll-off is 0.14 dB/µm (see Supplement 1).A typical A-scan measured in the experiment is shown in Fig. 2(B).To perform simultaneous measurements of depth positions in the tissue and Brillouin spectra, the cameras in the Brillouin and OCT spectrometers were triggered with the same trigger signal generated by a DAQ board controlled by the LabView program.

Measurement procedure
During the measurements, a volunteer placed the hand on the examination bed in a naturally flexed position without exerting any force.The laser light illuminated the wrist or palm of the hand.All the methods carried out in this work are under relevant guidelines and regulations from the local institutional review board (Ethikkommission der Friedrich-Alexander Universität, Erlangen, Germany).Testing on hand was performed as a self-test on the author of the manuscript who signed informed consent to participate.No other human experiments were performed in this work.
The manually controlled translation stage was used to adjust the position of the examination bed and to control the distance between the skin surface and the imaging lens.When the skin surface reached the focal position, a slow depth scan was performed.It is important to note that along with the controllable change in depth of the measured points, uncontrollable movements of the hand occurred due to breathing and heartbeat.The sample was displaced by several hundred micrometers within a few seconds.Without OCT tracking it is not possible to know the depth at which the Brillouin measurement is made.OCT tracking is used to correctly map the Brillouin spectra to the depth of the skin.The dependence of the Brillouin shift on the number of measurements before OCT-based correction is shown in Supplement 1.
For Brillouin spectra measurements, the acquisition time was 0.2 s, the sampling rate was 2 Hz, and the 0.3 s gap between spectral measurements was used by the spectrometer to process and store the data.The OCT imaging was triggered with the same trigger as the Brillouin spectra with a trigger rate of 2 Hz.The OCT measurement was performed with a sampling rate of 25 kHz and the duration of each measurement was 0.2 seconds, resulting in 5000 A-scans for one Brillouin spectrum.Typical A-scan and typical Brillouin spectra from a single measurement are shown in Fig. 2(B) and Fig. 2(C), respectively.A preliminary calibration with the mirror instead of the sample was used to determine the position of the focal point on the A-scan.The measurement depth was obtained as the distance between the position of the skin surface in the A-scan and the focal point (see Fig. 2(B)).The depth of position was then determined for each A-scan, and the mean depth position and standard deviation of depth were calculated.Because the peaks in the Brillouin spectra have low amplitude and are affected by noise, averaging is required to obtain the Brillouin shift value with high resolution, especially for a depth greater than 50 µm.The total measurement time was 5 minutes, during the first minute the sample position was adjusted to the focal position, and then the depth scanning was performed for 4 minutes to obtain enough measurement points for statistical averaging.From the measured data set, the spectra with the same depth positions were averaged.Due to the random motion of the sample, a different number of scans were measured for different depth positions.The minimum number of 3 and the maximum number of 12 spectra were used for averaging for a single point in depth.
No discomfort, erythema or photodamage was observed during or after the measurements.It is important to note that the sample was in constant motion during the measurement process due to the subject's breathing and heartbeat.Using OCT tracking, we found that the average standard deviation of the skin surface position during the 0.2-second acquisition time was equal to 24 µm.The motion of the sample during measurements limits the depth resolution of the current biomechanical testing technique for in vivo measurement to 24 µm, but prevents heating due to the constant movement of the focal point in the tissue.
Before in vivo measurements on human skin, the performance of the setup was tested ex vivo on layered agar phantoms and chicken skin samples.The results of the ex-vivo tests are available in the Supplement 1.

Results
The Brillouin spectra for the skin measured at the wrist are shown in Fig. 3(A).We observed two peaks in the Brillouin spectra at a depth within the first 60 µm of skin.The first peak in the Brillouin spectrum has a position near 7 GHz, while the second peak is centered near 9-10 GHz.Only a single peak is clearly visible in the spectra obtained for a depth greater than 60 µm.To analyze the measured spectra, we performed the fitting with the sum of two Lorentzian peaks: where ϑ is the frequency shift, ∆ B1 and ∆ B2 are the positions of the first and the second peaks in the Brillouin spectra, a 1 , a 2 , c 1 and c 2 are the fitting parameters.Only a part of the spectra with anti-Stokes peaks (from 4 to 14 GHz shift) was used for fitting.The result of the fitting is shown in Fig. 3(A), the blue and purple curves show the contributions of the first peak, I 1 (ϑ), and second peak, I 2 (ϑ), separately.The peak amplitudes were calculated using the following expression: I i (ϑ = ∆ Bi ) = a i /c i , where i is the number of a peak.The amplitudes of both peaks as a function of depth in the skin are shown in Fig. 3(C).The amplitude of the first peak has a maximum at the depth of 25 µm and decreases slowly with and second peak,  2 (), separately.The peak amplitudes were calculated using the following expression:  ( = ∆  ) =   /  , where  is the number of a peak.A typical Brillouin spectrum for the skin measured on the palm is shown in Fig. 3(D).For the palm measurement, only one peak was observed at a frequency of around 10 GHz.The intensity of this peak was much lower compared to the Brillouin peaks measured on the wrist (see Fig. 3(F)).This could be explained by the high roughness of the palm skin surface, which disturbs the focus.The value of the Brillouin shift varies from 9.5 to 10.5 GHz in the measured depth range up to 150 μm.

Discussion
The Brillouin spectra of human skin show different characteristics depending on the depth position in the case of measurements at the wrist.Two peaks were observed at the epidermal layer (which is approximately 80 μm deep at the wrist [27]).The second peak with the higher frequency appears in the upper 60 μm of the epidermis and we believe that it corresponds to the keratinized cells in the stratum corneum.Two peaks may appear simultaneously due to the natural movements of the hand during the Brillouin spectrometer signal acquisition.From OCT scanning, we found that the position of the skin surface has a standard deviation of 24 A typical Brillouin spectrum for the skin measured on the palm is shown in Fig. 3(D).For the palm measurement, only one peak was observed at a frequency of around 10 GHz.The intensity of this peak was much lower compared to the Brillouin peaks measured on the wrist (see Fig. 3(F)).This could be explained by the high roughness of the palm skin surface, which disturbs the focus.The value of the Brillouin shift varies from 9.5 to 10.5 GHz in the measured depth range up to 150 µm.

Discussion
The Brillouin spectra of human skin show different characteristics depending on the depth position in the case of measurements at the wrist.Two peaks were observed at the epidermal layer (which is approximately 80 µm deep at the wrist [27]).The second peak with the higher frequency appears in the upper 60 µm of the epidermis and we believe that it corresponds to the keratinized cells in the stratum corneum.Two peaks may appear simultaneously due to the natural movements of the hand during the Brillouin spectrometer signal acquisition.From OCT scanning, we found that the position of the skin surface has a standard deviation of 24 µm for 0.2 seconds, which means that the Brilllouin signal is collected from a trajectory that may cross many cells.The Brillouin shift of the first peak is higher near the surface and decreases significantly in the dermis layer.This is attributed to the fact that the dermis is softer than the epidermis, which is in agreement with the results obtained previously using optical coherence elastography [11].Tissue hydration is known to contribute to the Brillouin shift and elasticity [28,29], and the higher Brillouin shift in the epidermis may be due to the lower hydration of the outer skin layer.For the palm measurements, we observed only a single peak at 9.5-10.5GHz.This also supports the assumption that the peak at ∼ 10 GHz corresponds to the keratinized cells in the stratum corneum.In the palm area, the stratum corneum is approximately 170 µm [30], and in our measurement, we only reached the stratum corneum layer.The Brillouin shift for keratinized cells is higher because these cells are harder to protect the skin from the harsh environment [3].
In future work, Brillouin imaging of ex vivo human skin combined with additional fluorescence imaging or Raman spectroscopy may shed light on the origin of the two peaks observed in the Brillouin spectra.Further investigation of the influence of skin hydration and age is potentially promising because the ratio of the intensities of the two peaks as well as the values of the Brillouin shifts can be used as diagnostic parameters for the evaluation of skin condition.
One of the directions for future work is to develop a handheld probe for biomechanical testing of skin at different locations on the human body.A simple approach could be the development of a handheld probe with integrated polarizing beam splitter and quarter wave plate and collimators, although this handheld probe will not be very compact.For a compact solution, the implementation of a fiber-based endoscopic imaging probe would be more preferable.The probe can be fabricated using a hollow core fiber [31,32] or two different fibers for illumination and scattered light collection [33].
Due to the low efficiency of spontaneous Brillouin scattering and the low signal at the Brillouin peaks, the signal collection time was about 4 minutes.It was necessary to collect enough points for averaging for each depth to accurately detect the Brillouin shift.Also, due to the random movement of the hand during the in vivo test, many measurement points were excluded from the analysis because the sample was not properly positioned.In the future development of the technique, the measurement time can be reduced by increasing the sensitivity of the technique.The stimulated Brillouin scattering is a promising approach, which provides higher sensitivity with lower light power or shorter acquisition time [34][35][36].

Conclusion
We performed in vivo biomechanical testing of human skin using a combination of Brillouin microscopy and OCT sample position tracking.We observed that the Brillouin spectra of the epidermis and dermis of human skin in vivo have different characteristics.The Brillouin spectrum of the epidermis shows two peaks that may contribute to 2 types of cells: living cells and keratinized cells in the stratum corneum.The Brillouin spectrum for the dermis shows only one peak with a lower frequency shift.These results are consistent with the dermis being softer than the epidermis.We have shown that Brillouin imaging is a promising tool for non-invasive all-optical investigation of the biomechanics of different skin layers.In the future, this technique may be used for clinical applications to study the influence of pathological conditions on the elastic properties of the skin.

Fig. 1 .
Fig. 1.Schematic of depth-resolved biomechanical testing of human skin in vivo.

Fig. 2 .
Fig. 2. (A) Schematic of the experimental setup of Brillouin microscopy, combined with OCT for motion-corrected biomechanical testing of skin in vivo.SLD-super luminescent diode, BS -fiber-based beam splitter, PBS -polarization beam splitter, DM -dichroic mirror, QWP and HWP -quarter and half waveplates.Typical A-scan (B) and Brillouin spectra (C), measured in the experiment.

Fig. 3 .
Fig. 3. (A) The Brillouin spectra of skin measured on the wrist for different depths.The blue curve corresponds to  1() part and the violet curve corresponds to  2() part of the fitting with Eq.1.The dependence of 2 Brillouin peak positions (B) and intensities (C) on depth, measured on the wrist.(D) Typical Brillouin spectra of the skin, measured on the palm of the hand.The dependencies of Brillouin peak position (B) and intensity (C) on depth, measured on a palm.The amplitudes of both peaks as a function of depth in the skin are shown in Fig.3(C).The amplitude of the first peak has a maximum at the depth of 25 μm and decreases slowly with depth.At the depth position of 180 μm, it becomes comparable with the noise, and the first peak cannot be detected at greater depths.The amplitude of the second peak decreases sharply at the depth of 60 μm and the second peak is not visible at greater depths.The center positions of the Brillouin shift peaks (∆ 1 and ∆ 2 ) as a function of depth are shown in Fig.3(B).The center position of the first peak is approximately 7.05±0.05GHz for a depth of less than 60 μm.At a depth greater than 60 μm, the peak position slowly decreases with depth and reaches the value of 6.8±0.05GHz at the depth of 170 μm (see inset of Fig.3(B)).The position of the second peak varies between 9 and 10 GHz.

Fig. 3 .
Fig. 3. (A) The Brillouin spectra of skin measured on the wrist for different depths.The blue curve corresponds to I 1 (ϑ) part and the violet curve corresponds to I 2 (ϑ) part of the fitting with Eq.1.The dependence of 2 Brillouin peak positions (B) and intensities (C) on depth, measured on the wrist.(D) Typical Brillouin spectra of the skin, measured on the palm of the hand.The dependencies of Brillouin peak position (B) and intensity (C) on depth, measured on a palm.