Estimation of articular cartilage properties using multivariate analysis of optical coherence tomography signal

Raman Spectroscopy and Machine Learning Enables Estimation of Articular Cartilage Structural, Compositional, and Functional Properties

OBJECTIVE

To differentiate healthy from artificially degraded articular cartilage and estimate its structural, compositional, and functional properties using Raman spectroscopy (RS).

DESIGN

Visually normal bovine patellae (n = 12) were used in this study. Osteochondral plugs (n = 60) were prepared and artificially degraded either enzymatically (via Collagenase D or Trypsin) or mechanically (via impact loading or surface abrasion) to induce mild to severe cartilage damage; additionally, control plugs were prepared (n = 12). Raman spectra were acquired from the samples before and after artificial degradation. Afterwards, reference biomechanical properties, proteoglycan (PG) content, collagen orientation, and zonal (%) thickness of the samples were measured. Machine learning models (classifiers and regressors) were then developed to discriminate healthy from degraded cartilage based on their Raman spectra and to predict the aforementioned reference properties.

RESULTS

The classifiers accurately categorized healthy and degraded samples (accuracy = 86%), and successfully discerned moderate from severely degraded samples (accuracy = 90%). On the other hand, the regression models estimated cartilage biomechanical properties with reasonable error (≤ 24%), with the lowest error observed in the prediction of instantaneous modulus (12%). With zonal properties, the lowest prediction errors were observed in the deep zone, i.e., PG content (14%), collagen orientation (29%), and zonal thickness (9%).

CONCLUSION

RS is capable of discriminating between healthy and damaged cartilage, and can estimate tissue properties with reasonable errors. These findings demonstrate the clinical potential of RS.

Resolving the Near-Infrared Spectrum of Articular Cartilage

OBJECTIVE

Spectroscopic techniques, such as near-infrared (NIR) spectroscopy, are gaining significant research interest for characterizing connective tissues, particularly articular cartilage, because there is still a largely unmet need for rapid, accurate and objective methods for assessing tissue integrity in real-time during arthroscopic surgery. This study aims to identify the NIR spectral range that is optimal for characterizing cartilage integrity by (a) identifying the contribution of its major constituents (collagen and proteoglycans) to its overall spectrum using proxy constituent models and (b) determining constituent-specific spectral contributions that can be used for assessment of cartilage in its physiological state.

DESIGN

The NIR spectra of cartilage matrix constituent models were measured and compared with specific molecular components of organic compounds in the NIR spectral range in order to identify their bands and molecular assignments. To verify the identified bands, spectra of the model compounds were compared with those of native cartilage. Since water obscures some bands in the NIR range, spectral measurements of the native cartilage were conducted under conditions of decreasing water content to amplify features of the solid matrix components. The identified spectral bands were then compared and examined in the resulting spectra of the intact cartilage samples.

RESULTS

As water was progressively eliminated from cartilage, the specific contribution of the different matrix components was observed to correspond with those identified from the proxy cartilage component models.

CONCLUSION

Spectral peaks in the regions 5500 to 6250 cm-1 and 8100 to 8600 cm-1 were identified to be effective for characterizing cartilage proteoglycan and collagen contents, respectively.

Articular cartilage optical properties in the near-infrared (NIR) spectral range vary with depth and tissue integrity

Optical properties of biological tissues in the NIR spectral range have demonstrated significant potential for in vivo diagnostic applications and are critical parameters for modelling light interaction in biological tissues. This study aims to investigate the optical properties of articular cartilage as a function of tissue depth and integrity. The results suggest consistent wavelength-dependent variation in optical properties between cartilage depth-wise zones, as well as between healthy and degenerated tissue. Also, statistically significant differences (p<0.05) in both optical properties were observed between the different cartilage depth-wise zones and as a result of tissue degeneration. When taken into account, the outcome of this study could enable accurate modelling of light interaction in cartilage matrix and could provide useful diagnostic information on cartilage integrity.

Two- and three-dimensional optical coherence tomography to differentiate degenerative changes in a rat meniscectomy model

Optical coherence tomography (OCT) is an attractive tool for evaluating cartilage. We developed an OCT system that reconstructs and analyzes a three-dimensional (3D) OCT image by determining the cartilage surface and cartilage-bone boundary from the image taken with currently available OCT devices. We examined the usefulness of 3D renderings of OCT images. In a rat meniscectomized model, the tibia was harvested after 0, 2, 4, or 8 weeks (n = 6). We scanned 300 slices in the y-plane to cover a 4 × 3 × 6-mm section (x-plane; 10 µm × 400 pixels, y-plane; 10 µm × 300 pixels, z-plane; 12.66 µm × 500 pixels) of the medial tibial cartilage. The cartilage surface line and the cartilage-bone boundary were plotted semi-automatically. Slices from 300 two-dimensional (2D) sequential images were systematically and visually checked and corrected, as necessary. We set a region of interest in the cartilage and quantified the cartilage volume in the 3D image. The Osteoarthritis Research Society International (OARSI) histological score was also obtained. The cartilage volume determined using 3D OCT images was 0.291 ± 0.022 mm³ in the normal, 0.264 ± 0.009 mm³ at 2 weeks, 0.210 ± 0.012 mm³ at 4 weeks, and 0.205 ± 0.011 mm³ at 8 weeks. The cartilage volume significantly decreased at 4 and 8 weeks and was significantly correlated with the OARSI histological score (r = -0.674; P = .002). Although the 3D image information could be obtained from the 2D images, the 3D OCT images provided easier-to-understand information because the 3D reconstructed cartilage provided information about the smoothness of the surface, the area, and depth of the defect at a glance.

Predictive evaluation of powder X-ray diffractograms of pharmaceutical formulation powders based on infrared spectroscopy

BACKGROUND

To ensure quality and stability, monitoring systems are recommended to analyze pharmaceutical manufacturing processes.

OBJECTIVE

This study was performed to predict powder X-ray diffraction (PXRD) patterns of formulation powders through attenuated total reflectance (ATR)-infrared (IR) spectroscopy in a nondestructive manner along with chemometrics.

RESULTS

Caffeine anhydrate, acetaminophen, and lactose monohydrate were grinded at six weight ratios. The six sample groups were evaluated using ATR-IR spectroscopy and PXRD analysis. Partial least squares models were constructed to predict the PXRD intensities of the samples from the ATR-IR spectra. The prediction accuracy on the prepared PLS regression models was as high as R2 = 0.993.

CONCLUSIONS

Linear relationships were obtained between the prediction data set and reference PXRD intensity at each degree. 2D PLS regression coefficient analysis enabled the analysis of the correlation between PXRD patterns and IR spectra.

Optical spectroscopic characterization of human meniscus biomechanical properties

This study investigates the capacity of optical spectroscopy in the visible (VIS) and near-infrared (NIR) spectral ranges for estimating the biomechanical properties of human meniscus. Seventy-two samples obtained from the anterior, central, and posterior locations of the medial and lateral menisci of 12 human cadaver joints were used. The samples were subjected to mechanical indentation, then traditional biomechanical parameters (equilibrium and dynamic moduli) were calculated. In addition, strain-dependent fibril network modulus and permeability strain-dependency coefficient were determined via finite-element modeling. Subsequently, absorption spectra were acquired from each location in the VIS (400 to 750 nm) and NIR (750 to 1100 nm) spectral ranges. Partial least squares regression, combined with spectral preprocessing and transformation, was then used to investigate the relationship between the biomechanical properties and spectral response. The NIR spectral region was observed to be optimal for model development (83.0%≤R2≤90.8%). The percentage error of the models are: Eeq (7.1%), Edyn (9.6%), Eϵ (8.4%), and Mk (8.9%). Thus, we conclude that optical spectroscopy in the NIR range is a potential method for rapid and nondestructive evaluation of human meniscus functional integrity and health in real time during arthroscopic surgery.

Monitoring osteoarthritis progression using near infrared (NIR) spectroscopy

We demonstrate in this study the potential of near infrared (NIR) spectroscopy as a tool for monitoring progression of cartilage degeneration in an animal model. Osteoarthritic degeneration was artificially induced in one joint in laboratory rats, and the animals were sacrificed at four time points: 1, 2, 4, and 6 weeks (3 animals/week). NIR spectra were acquired from both (injured and intact) knees. Subsequently, the joint samples were subjected to histological evaluation and glycosaminoglycan (GAG) content analysis, to assess disease severity based on the Mankin scoring system and to determine proteoglycan loss, respectively. Multivariate spectral techniques were then employed for classification (principal component analysis and support vector machines) and prediction (partial least squares regression) of the samples’ Mankin scores and GAG content from their NIR spectra. Our results demonstrate that NIR spectroscopy is sensitive to degenerative changes in articular cartilage, and is capable of distinguishing between mild (weeks 1&2; Mankin <=2) and advanced (weeks 4&6; Mankin =>3) cartilage degeneration. In addition, the spectral data contains information that enables estimation of the tissue’s Mankin score (error = 12.6%, R² = 86.2%) and GAG content (error = 7.6%, R² = 95%). We conclude that NIR spectroscopy is a viable tool for assessing cartilage degeneration post-injury, such as, post-traumatic osteoarthritis.

Characterization of Articular Cartilage Recovery and Its Correlation with Optical Response in the Near-Infrared Spectral Range

OBJECTIVES

In this study, we examine the capacity of a new parameter, based on the recovery response of articular cartilage, to distinguish between healthy and damaged tissues. We also investigate whether or not this new parameter correlates with the near-infrared (NIR) optical response of articular cartilage.

DESIGN

Normal and artificially degenerated (proteoglycan-depleted) bovine cartilage samples were nondestructively probed using NIR spectroscopy. Subsequently they were subjected to a load and unloading protocol, and the recovery response was logged during unloading. The recovery parameter, elastic rebound ( E_R), is based on the strain energy released as the samples underwent instantaneous elastic recovery.

RESULTS

Our results reveal positive relationship between the rebound parameter and cartilage proteoglycan content (normal samples: 2.20 ± 0.10 N mm; proteoglycan-depleted samples: 0.50 ± 0.04 N mm for 1 hour of enzymatic treatment and 0.13 ± 0.02 N mm for 4 hours of enzymatic treatment). In addition, multivariate analysis using partial least squares regression was employed to investigate the relationship between E_R and NIR spectral data. The results reveal significantly high correlation ( R²_cal = 98.35% and R²_val = 79.87%; P < 0.0001), with relatively low error (14%), between the recovery and optical response of cartilage in the combined NIR regions 5,450 to 6,100 cm^-1 and 7,500 to 12,500 cm^-1.

CONCLUSION

We conclude that E_R can indicate the mechanical condition and state of health of articular cartilage. The correlation of E_R with cartilage optical response in the NIR range could facilitate real-time evaluation of the tissue's integrity during arthroscopic surgery and could also provide an important tool for cartilage assessment in tissue engineering and regeneration research.

Porosity predicted from ultrasound backscatter using multivariate analysis can improve accuracy of cortical bone thickness assessment

A rapidly growing area of interest in quantitative ultrasound assessment of bone is to determine cortical bone porosity from ultrasound backscatter. Current backscatter analyses are based on numerical simulations, while there are no published reports of successful experimental measurements. In this study, multivariate analysis is applied to ultrasound reflections and backscatter to predict cortical bone porosity. The porosity is then applied to estimate cortical bone radial speed of sound (SOS) and thickness using ultrasound backscatter signals obtained at 2.25 and 5 MHz center frequencies from cortical bone samples (n = 43) extracted from femoral diaphyses. The study shows that the partial least squares regression technique could be employed to successfully predict (R²= 0.71-0.73) cortical porosity. It is found that this multivariate approach can reduce uncertainty in pulse-echo assessment of cortical bone thickness from 0.220 to 0.045 mm when porosity based radial SOS was applied, instead of a constant value from literature. Upon further validation, accurate estimation of cortical bone porosity and thickness may be applied as a financially viable option for fracture risk assessment of individuals.

Polarized light interaction with tissues

This tutorial-review introduces the fundamentals of polarized light interaction with biological tissues and presents some of the recent key polarization optical methods that have made possible the quantitative studies essential for biomedical diagnostics. Tissue structures and the corresponding models showing linear and circular birefringence, dichroism, and chirality are analyzed. As the basis for a quantitative description of the interaction of polarized light with tissues, the theory of polarization transfer in a random medium is used. This theory employs the modified transfer equation for Stokes parameters to predict the polarization properties of single- and multiple-scattered optical fields. The near-order of scatterers in tissues is accounted for to provide an adequate description of tissue polarization properties. Biomedical diagnostic techniques based on polarized light detection, including polarization imaging and spectroscopy, amplitude and intensity light scattering matrix measurements, and polarization-sensitive optical coherence tomography are described. Examples of biomedical applications of these techniques for early diagnostics of cataracts, detection of precancer, and prediction of skin disease are presented. The substantial reduction of light scattering multiplicity at tissue optical clearing that leads to a lesser influence of scattering on the measured intrinsic polarization properties of the tissue and allows for more precise quantification of these properties is demonstrated.