1. Introduction

Articular cartilage (AC) is a specialized connective tissue with the main function of reducing the frictional forces during the movement in the joint and providing the resistance to the shock loads. Remodeling processes take place in articular cartilage. They include degradation and synthesis of the major components (collagen and proteoglycans (PG)) of the extracellular matrix (ECM) that forms the basis of the cartilage tissue. The ECM consists mainly of the water (65-80%), collagen (10-30%), PGs (3-10%) and a relatively small number of cells and chondrocytes [1]. Several other classes of molecules can be found in the ECM in smaller amounts, not exceeding one percent. These include lipids, phospholipids, noncollagenized proteins [2,3]. Articular cartilage is limited in its regenerative ability and constantly degenerates, which leads to the osteoarthritis (OA). OA is a chronic arthropathy characterized by damage and destruction of the articular cartilage, which leads to the changes in the ECM. Degradation of the ECM is associated with disruption of the collagen network, loss of PG content, and increase in water content [4], combined with other joint changes, including the development of the osteophytes. Articular cartilage is hyaline cartilage with a thickness from 1 to 4 mm, that has a heterogeneous structure and several characteristic zones. Depending on the distance from the cartilage surface, several characteristic zones can be distinguished by their functions and structure. The surface (tangential) zone occupies about 10% of the cartilage volume. The collagen fibers of this zone are tightly packed and aligned parallel to the articular surface. The middle (transition) zone is up to 15% of the total cartilage volume, and it contains PGs and thicker collagen fibrils. In this zone, collagen is organized obliquely. A deep zone occupies approximately 70-90% of the articular cartilage volume, contains collagen fibrils of the largest diameter in a radial arrangement, has the highest PG content and lowest water concentration of about 65% [5–7]. This zoning is most typical for the extracellular matrix of articular cartilage, although there are differences in both layer thicknesses and water/collagen component ratios [8,9]. Collagen is the main component of cartilage and is about 60% of the dry mass [5].

A particular interest in the diagnosis of early OA is the determination of the water/collagen ratio, the imbalance of which is the first factor in the progression of OA [4,10]. One way to diagnose and evaluate the main components is through the optical spectroscopy techniques, especially in the IR range. With an increase in the wavelength in the IR region, in general, a decrease in the scattering and an increase in the absorption of water follow. At the same time, the absorption of water has a local minima, which makes it possible to observe the absorption of other components and, in particular, collagen, the relative content of which is of interest.

The NIR region (700-2500 nm) is actively used in research work to solve diagnostic problems. The application of NIR-I (700-1000 nm) is primarily related to several characteristic overtones of the molecular vibrations of the water, lipids, and different modifications of the hemoglobin. NIR-I has found applications in breast cancer diagnosis [11,12]. NIR-II (1000-1350 nm) and NIR-III (1550-1870 nm) regions were used in the tasks of the water and lipid content estimation. Diffuse reflectance spectroscopy in these areas has shown the possibility of providing an accurate classification of the affected tissue. For example, this approach has been used to distinguish vulnerable atherosclerotic plaques from stable plaques [13], for distinguishing different types of breast cancer from nonmalignant breast neoplasms [14], to study the structure and composition of the cartilage joints depending on the degree of degradation, using bovine articular cartilage as an example [15] and for assessing the water content in skin edema [16]. Applicability of the IV transparency window or SWIR-II (2100-2350 nm) in clinical medicine is of significant interest for different tissues, including cartilage, blood vessels and skin [15,17,18]. Interest in this area is due to the presence of two collagen absorption peaks at 2170 nm and 2350 nm and two lipid absorption peaks at 2180 nm and 2350 nm [19–21].

There are a limited amount of the radiation sources in the fourth transparency window. Among them it is worth to distinguish the supercontinuum sources, which, however, possess low power (less than 10mW) in the required range. Optical parametric oscillators are also used, the main disadvantage of which is the complexity and high cost of installation. Among fiber laser sources can be distinguished holmium lasers, but their maximum wavelength of operation is 2.17 µm [22]. A disadvantage of commercially available laser diodes is the availability of models with only a specific set of wavelengths (2160, 2180, 2200, 2230 nm). Following these only a limited number of studies were primarily focused on the transmission and absorption characteristics in SWIR-II region [20,23]. OCT systems at the wavelength of 2.1 µm are also currently under development [24]. One of the promising solutions is the use of the lasers based on A$_2$B$_6$ group chalcogenides doped with chromium ions, such as Cr$^{2+}$:ZnSe [25–32]. The uniqueness of these active media lies in the possibility of ultra-wide wavelength tuning from 1800 to 3500 nm, which allows them to be used for a wide range of applications. The possibility of using such lasers as a source for spectroscopy, OCT, and optoacoustic imaging makes it possible to significantly improve the diagnosis of the biotissues in vivo. In this paper, we present the initial results of diffuse reflectance spectroscopy of cartilage tissue and its optical phantom in the fourth transparency window using a Cr$^{2+}$:ZnSe tunable laser.

2. Methods and materials

2.1 Experimental setup

To investigate the diffuse reflectance of the samples, the setup represented in Fig. 1 was used. The laser, based on a Cr$^{2+}$:ZnSe active medium and tunable in the range from 2.1 to 2.4 µm, was used as a radiation source. The laser beam was coupled by the lens collimator (F810SMA-2000, Thorlabs, USA) into the illumination multimode silica fiber (IPG Photonics, Russia) with a core diameter of 550 µm and NA 0.22 and transmitted to the sample. A second identical optical fiber (collection fiber) was attached to the movable stage and used to collect the diffuse reflectance signal from the sample and deliver it to a spectrometer covering the wavelength range from 1.0 to 2.5 µm with 9.5 nm resolution (Avantes NIR-256, Netherlands).

 

Fig. 1. a) Dual fiber probe scheme. b) Experimental setup for diffuse reflectance spectroscopy. c) The spectra of the samples’ main components.

Download Full Size | PDF

A polytetrafluoroethylene (PTFE) sample was used for measuring the reference spectra $I_\textrm {ref}$ to take into account the spectral dependence of the laser intensity, the sensitivity of the spectrometer, and losses along the entire optical path. The optical characteristics of PTFE allow it to be used as a reference material in this range [33]. For the reference spectra, both fibers were set 10 mm above the PTFE sample and were in contact with each other during the calibration process throughout the experiment. For the sample spectra measurement both fibers were in contact with the specimen during the measurement. Distance between the fibers and corresponding penetration depth was varied by moving the collection fiber relative to the illumination fiber using a mechanical system based on the manual linear translation stage with 0.01 mm accuracy. The fibers were set at a given distance, and then a wavelength scan was performed in the range from 2.1 to 2.4 µm producing the diffuse reflectance spectrum. The reference signal was not recalibrated for different fibre spacing, as no change in the shape of the spectrum was observed.

2.2 Tunable laser source

The chromium-doped ZnSe monocrystal was used as an active media due to its broadband tuning range capability. To achieve proper tuning range, it is necessary to select a proper tuning method.

The diffuse reflectance spectroscopy method implies the illumination radiation delivery through the optical fiber. In this case, the laser beam must not change its propagation direction during tuning process which makes it a limiting factor. Commonly used methods such as a dispersive prism or diffracting grating in Littrow or Littman configurations either do not allow to save the propagation direction of the output laser beam or lead to the large intracavity losses [34,35].

Following this, the most suitable tuning method is to use a Lyot filter consisting of one or several birefringent plates (BFP). To obtain broadband tuning from 2.1 to 2.4 µm the appropriate Lyot filter parameters must be calculated. Existing models either do not take into account all of these parameters [36] or have typos [37]. In this paper these problems were taken into account. Compared to work [37] wrong matrix multiplication order (Eq. (3)) and typos in the main equation for the BFP transmission matrix (Eq. (5)) were corrected.

Presented model based on a Jones matrix formalism, thus each optical element is represeted by 2x2 transmission matrix. Eq. (1) shows such a matrix for BFP.

To determine the transmission spectrum of the resonator cavity, the squares of the maximum eigenvalues absolute values of the round trip matrix must be calculated.

Using this model, a Lyot filter consisting of MgF$_{2}$ BFP with 0.55 mm thickness and crystal axis laying on the surface of the plate was calculated. Two additional ZnSe Brewster plates were set into the resonator cavity to increase the Lyot filters modulation depth, as represented in Fig. 2.

 

Fig. 2. Theoretical transmission spectrum of the Lyot filter for fixed values of the angle between the normal to the surface and the optical axis of the BFP. The two edge positions at the 45.1$^\circ$ and 69.5$^\circ$ refer to the best Lyot filter transmission at the 2.1 and 2.4 µm respectively.

Download Full Size | PDF

CW Tm-fiber laser with 1908 nm central wavelength and maximum output power of 20 W was used as a pump source. The resonator was built using a mirror-lens scheme to increase the cavity length with the following capability to place a Lyot filter inside of it, as represented in Fig. 3. The resonator consists of an input meniscus with a 60 mm curvature radius, flat output coupler with 30% transmission in a spectral range from 2.1 to 2.5 µm and intracavity CaF$_{2}$ lens with 50 mm focal length. Cr$^{2}$:ZnSe monocrystal (3.2 x 5.7 x 2.8 mm and $1.1\cdot 10^{19}$ cm$^{-3}$ chromium ion concentration) was used as an active medium.

 

Fig. 3. Scheme of the tunable Cr$^{2+}$:ZnSe laser.

Download Full Size | PDF

2.3 Materials

To evaluate the applicability of diffuse reflectance spectroscopy in assessing the water and collagen content, two types of samples were used: gelatin optical phantoms and cartilage tissue with varying water content. For gelatin phantoms, water and collagen relative content was determined by the mass of the components: water and gelatin. For cartilage samples, the change in water content was carried out by drying.

2.3.1 Optical phantoms

Animal gelatin (containing collagen as a natural absorber), water and TiO$_2$ particles as a diffuser (RusChem, St. Petersburg, Russia) were used for phantom preparation. The design technique was as follows. Gelatin solutions in water with concentrations in the range of 10-50% of the total mass were prepared. The selected range corresponds to the amount of the collagen in the cartilage tissue [38]. Then fine particles of pre-crushed TiO$_2$ powder were added to the solution at a concentration of 0.1% of the total mass and the resulting solution was mixed using a high-pressure homogenizer (Mef91, Melfiz, Russia). The mixing process took no more than two minutes for each sample. Next, the solution was poured into Petri dishes and hardened at 5$^\circ$C. The approximate thickness of the samples was 6 mm. The amount of titanium dioxide was chosen in such a way that the reflected signal from the phantoms corresponded to the signal from the cartilage tissue. Thus, the phantoms had values of scattering and absorption coefficients typical for cartilage tissue ($\mu _s^{'} = 1~\mathrm {mm}^{-1}$, $\mu _a = 1.3~\mathrm {mm}^{-1}$) [39].

2.3.2 Cartilage samples

Cartilage tissue samples (six in total) were extracted from the bovine knee joints of the animals aged 16-24 months obtained at a local slaughterhouse within 24 hours after slaughter. Cartilaginous specimens without any signs of tissue destruction on the surface were used in the experiment. Small slices of cartilage tissue from 1 mm to 2 mm thick on subchondral bone was used for the measurements. The bone occupied 30% of the total weight and was cut out along with the cartilage tissue samples to avoid the contribution of the Petri dish substrate. Tissue specimens were stored in a sodium chloride physiological solution to prevent premature drying. Diffuse reflectance spectra (DRS) spectra were measured during drying at room temperature (24$^\circ$C) for different distances between illumination/collection fibers.

Cartilage tissue spectra were registered at time intervals of 5-15 minutes for 2.5 hours to observe spectral changes caused by the water evaporation. Cartilage tissue samples were weighed every ten minutes for several hours, with simultaneous measurements of spectral data. The average measurement time for the single sample was one minute. To calculate the dry mass, the cartilage samples were dried at room temperature of 24$^\circ$C for 30 hours so that their mass stopped changing.

2.4 Spectra measurements

Diffuse reflectance intensity spectra were obtained for different wavelengths of tunable laser in the following way. For every filter position, detected spectra were background corrected by subtracting the mean value in 2000-2050 nm range, then integrated in a window of 40 nm around the maximum position to obtain reflection intensity value. The corresponding laser wavelength was obtained as a maximum position in the registered spectra. All wavelength and intensity value pairs obtained during the scan were joined and linearly interpolated on a 2140-2320 nm range with a hundred nodes resulting in intensity spectra $I(\lambda )$. Each wavelength scan was repeated three times. The obtained spectra were then smoothed by the Savitzky-Golay (S-G) filter with a polynomial degree of 3 and a window size of 40 and were averaged over the three measurements. For each sample measurement the reference spectrum $I_\mathrm {ref}(\lambda )$ from a PTFE plate was measured.

The reflectance spectra were calculated as

Diffuse reflectance spectra contains information about the chromophores concentration, however the problem of determination of total absorption spectra and their contribution does not have exact solution in the general case. The inability to measure the exact value of the optical path as well as the concentration of each component makes it necessary to use an optical density value that contains this information.

The optical density is calculated in this case as

Component concentration then can be estimated using differences of optical density at wavelengths corresponding to maximum and minimum of component absorption [40]. Similarly to this approach, to estimate the contribution of each chromophore, one can approximate the optical density spectrum by the sum of the absorption or optical density spectra of the components. In our case, the optical density of tissue and phantom spectra were fitted with as a linear combination of the optical density of the chromophores: water and collagen in case of tissue samples or gelatin in case of phantoms:

The optical density spectra of gelatin and collagen were obtained using the same DRS setup with the fibers located at a minimum distance. To assess the possibility of detecting the response from deeper layers of the sample, a series of measurements were carried out for different distances between the optical fibers, which corresponds to a different characteristic detection depth. Simulation of the process of propagation of light into the cartilage tissue showed that at a distance of 0.5 mm between the fibers, the average penetration depth is 0.22 mm. Since the absorption of water is quite high, with increasing distance, a significant attenuation of the signal was observed, and the distance between the fibers varied in limited the range. For optical phantoms this range is from 0.2 to 0.6 mm, and for cartilage tissues distance between the fibers was from 0.2 to 0.4 mm

3. Experiment results and discussion

In the current study, the possibilities of spectroscopy in the 4th transparency window for analyzing the variation in the relative water and collagen content in articular cartilage were investigated. The solid-state tunable laser source based on a Cr$^{2+}$:ZnSe crystal was developed for the diffuse reflectance spectroscopy system. The first objects to be considered were gelatin-based optical phantoms with optical properties and concentrations of the components (water, gelatin) corresponding to articular cartilage. Furthermore, SWIR-II spectroscopy enabled determination of the relative content of water and collagen in bovine knee cartilage specimens during their drying.

3.1 Tunable laser performance

The laser was continuously tuned by Lyot filter rotation in a plane perpendicular to the optical axis in the wavelength range from 2.1 to 2.4 µm with no change in the propagation direction. Spectral characteristics at the several wavelengths are shown in Fig. 4(a). Output power dependences on the absorbed pump power for different wavelength are shown in Fig. 4(b).

 

Fig. 4. a) Laser spectrum at several wavelengths with FWHM<15 nm. b) Laser efficiency and beam cross-section on the inset.

Download Full Size | PDF

The beam quality was measured right after the laser resonator using a knife-edge method. The laser beam quality parameter M$^{2}$ was 1.6. The cross-section of the output laser beam has astigmatism due to the presence of Brewster plates in the cavity, as shown on the inset in Fig. 4(b).

3.2 Changes in the optical properties of gelatin phantoms

Experimentally obtained spectra of optical density for a distance between two fibers of 0.6 mm (measured between inner surfaces) are represented in Fig. 5(a). As one can see, when the percentage of gelatin in the samples increases, there is a decrease in the contribution of H$_2$O spectrum (see Fig. 5(b)) and a more clear manifestation of collagen peaks at 2180 nm and 2280 nm which makes it possible to quantify their contributions. Each spectrum was fitted with Eq. (5). The weight factors as a function of the percentage of gelatin in the phantoms and their linear fit are represented in Fig. 5(c). It can be noticed that gelatin concentration correlates with the gelatin amplitude $A_\mathrm {gelatin}$, and negatively correlates with $A_\mathrm {water}$. Next, a series of experiments were performed to evaluate the behavior of the components according to the light penetration depth for various distances between the illumination/collection fibers. Linear fittings of the concentration dependencies of $A_\mathrm {gelatin}$ and $A_\mathrm {water}$ were established for 0.2 mm, 0.4 mm, and 0.6 mm distances. To examine the behavior of these dependencies, the slope coefficients of the linear approximation for the water component were compared in Fig. 5(d). The values of the slope coefficient for several distances lie within 10% difference, which indicates homogeneity of the optical properties of the phantom.

 

Fig. 5. a) Optical density spectra of gelatin-based optical phantoms at a distance of 0.6 mm between the illumination/collection fibers with the standard deviations for optical density values obtained for the three sample measurements. b) Fit of the OD spectrum for 40% gelatin optical phantom, with a combination of water (blue dashed line) and gelatin (green dashed line) spectra. c) Weight coefficients for water and gelatin components of optical phantoms as a function of gelatin concentration for spectra obtained at a distance of 0.6 mm between the illumination/collection fibers. The dash-dotted lines show linear trends of the weight coefficients for water ($R^{2}$ = 0.84) and gelatin ($R^{2}$ = 0.84) components. Vertical lines represent standard deviations for amplitude values obtained by the three sample measurements. d) Absolute values of the slope coefficient $K_{\textrm {water}}$ for the different distances between the illumination/collection fibers.

Download Full Size | PDF

Thus, by using the developed method it is possible to determine relative differences of water and gelatin content in optical phantoms with concentrations of gelatin in the range 10-50%. In our case, such changes in optical properties can be observed even at low concentrations of gelatin in optical phantoms, which indicates the method applicability to specimens with high water content.

3.3 Changes in the optical properties of cartilage tissue samples during drying

To investigate the possibility of observing changes in the water-collagen component in systems close to in situ measurements, cartilage tissue specimens were studied during the drying process. Measurements of the optical characteristics were carried out for different distances between the illumination and collection fibers. Consequently, the optical response of the signal from different depths was investigated.

Experimentally obtained spectra of optical density, fitting result, and weight coefficients for water and collagen content during drying are shown in Fig. 6(a-c). During the first two hours of drying active evaporation of free water takes place, and almost 50% of the total weight of articular cartilage is lost [42]. Since the samples of the articular cartilage on a subchondral bone were used for measurements in this case, in two hours the mass of the tissue samples decreased by 31.7%. A direct comparison of spectra during drying demonstrates that the absorption of collagen at 2170 nm and 2280 nm becomes more pronounced. The result of decomposing of the optical density spectra of cartilage tissue into water and collagen components is represented in Fig. 6(b). As can be seen, the shape of the collagen peaks in the measured spectrum and in the approximating one is somewhat different, which led to an increase in the approximation error and the restored collagen contribution. The change in the contribution of the collagen component from the beginning of the drying process to its end is approximately 28%, which is close to the change in the total mass of the cartilage sample (31%). This suggests the possibility of determining collagen content in in vivo experiments. At the same time, the error in the approximation of water in this way is less pronounced, as can be seen in Fig. 6(c). Linear fit of water component dependence on sample mass was carried out for distances 0.2 mm and 0.4 mm. The slope coefficient increased by almost 25% when the distance was increased, which may indicate a heterogeneous structure and a change in the amount of water in the sample depending on the penetration depth during drying.

 

Fig. 6. a) Optical density spectra of the cartilage tissue samples at a distance of 0.4 mm between the illumination/collection fibers during drying with the standard deviations for optical density values obtained by the three sample measurements. b) Fit of the OD spectrum for cartilage tissue after 33 minutes of drying with a combination of water (blue dashed line) and collagen (orange dashed line) spectra. c) Dependence of the weight coefficients for water and collagen components of cartilage tissue at a distance of 0.4 mm between the illumination/collection fibers on the sample weight during the drying process. The dash-dotted lines show linear trends of the weight coefficients for water ($R^{2}$ = 0.96) component. Vertical lines represent standard deviations for amplitude values obtained by the three sample measurements. d) Absolute values of the slope coefficient $K_{\textrm {water}}$ of the water amplitude dependence on cartilage mass (see panel c) for several distances between illumination/collection fibers.

Download Full Size | PDF

Obtained results demonstrate the possibility of determining the relative changes of the water content in the cartilage tissue, while the determination of the collagen component requires further improvement. This task seems to be possible due to the results obtained for the gelatin phantoms. Possible approaches could include expansion of the laser tuning region and improvement of the fitting procedure.

Current work represents the first results on evaluating the possibility of determining of the relative content of water and collagen in optical phantoms of cartilage tissue, as well as on evaluating the possibility of determining the relative content of water and collagen in cartilage tissue using optical fiber probe in a wavelength range corresponding to the fourth window of transparency. Although it is the first application of such a technique, there are a number of prerequisites that determine the interest in this kind of research. In particular, earlier works established that the water content in cartilage tissue for ex vivo cartilage samples can be determined using IR spectroscopy in the transmission configuration [42]. Measurement of the transmission spectra of pellets containing cartilage main components and their comparison with the spectra of ex vivo samples in this area indicated that local absorption maxima in the region of 2170 and 2280 nm correspond to the absorption maxima of collagen and proteoglycans [43]. The content of these substances determines the functional state of the cartilaginous tissue [44], which makes the tuning range presented interesting. Since it is impossible to carry out in situ measurements in transmission configuration, the measurement setups using optic fiber probes are in demand. It is known that using such configuration and diffuse reflection signal, it is possible to observe the characteristic peaks of the overtones of OH, CH and NH vibrations of cartilage tissue components, however, the use of broadband light sources allows to reliably register the response in the wavelengths range up to 1800 nm [45]. This limitation can be overcome using laser sources, as was shown in the present work. Nevertheless, measurements in the region up to 1800 nm make it possible, using statistical and machine learning techniques, to build models for predicting the concentrations of the main components of cartilage tissue during in vitro and in situ measurements [46], to evaluate its biomechanical properties [45], as well as to calculate indexes that are statistically significantly different for different grades according to the ICRS and MANKIN scores, allowing to classify the state of the tissue [47,48]. At the same time, it was demonstrated that the use of the wavelength region up to 1400 nm can lead to the observable contribution from the subchondral bone located under the cartilage [49], which can complicate the interpretation of the results. Thus, the observation of the response in the longer wavelength region of the spectrum considered in the present work is of great interest.

Further study of ex vivo and in situ samples in the fourth window of transparency, including the samples with the earliest degenerative changes, will be an important step for development of a diagnostic tool that can change the patient’s treatment strategy. Evaluation of both water and collagen content can be useful in detecting structural changes in cartilage, which may allow early stages of osteoarthritis progression to be detected.

4. Conclusions

The current study confirmed that diffuse reflectance spectroscopy in the SWIR-II range can be used to observe changes in the water and collagen content of articular cartilage using characteristic peaks of collagen absorption at 2170 nm and 2280 nm and water absorption in this region.

The broadband tuning range with a high emission cross-section of a Cr$^{2+}$:ZnSe crystal and the Lyot filter tuning method advantages proved the capability of the presented tunable laser for such a spectroscopic method.

Analysis of the diffuse reflectance spectra of gelatin-containing optical phantoms with a varying content of water and gelatin showed that using the absorption peaks of these substances, one can characterize their relative content using an observation scheme that can potentially be used for in situ measurements. It was also demonstrated that the observation of the response is possible in the range of distances up to 0.6 mm between the optical fibers, which may allow probing cartilage tissue in a certain depth range.

The results obtained for the tissue samples during the drying process showed the expected trend for a decrease in the water contribution, which indicates that it is fundamentally possible to use this method to measure water and collagen content in cartilage tissue samples. By the end of the drying process, the contribution of the collagen component increased by 28%, which corresponds to a real change in the mass of the cartilage tissue sample, which decreased by 31%. The study of DRS spectra depending on the distance between the fibers up to 0.4 mm showed the ability to register the signal response from different depths.

The obtained data suggest that this technique is promising for the study of the cartilage tissue diseases and also confirms the possibility of setting up new diagnostic systems in the fourth transparency window.