Near-infrared spectroscopy enables quantitative evaluation of human cartilage biomechanical properties during arthroscopy

Abstract

Objective: To investigate the feasibility of near-infrared (NIR) spectroscopy (NIRS) for evaluation of human articular cartilage biomechanical properties during arthroscopy. Design: A novel arthroscopic NIRS probe designed in our research group was utilized by an experienced orthopedic surgeon to measure NIR spectra from articular cartilage of human cadaver knee joints (ex vivo, n ¼ 18) at several measurement locations during an arthroscopic surgery. Osteochondral samples (n ¼ 265) were extracted from the measurement sites for reference analysis. NIR spectra were remeasured in a controlled laboratory environment (in vitro), after which the corresponding cartilage thickness and biomechanical properties were determined. Hybrid multivariate regression models based on principal component analysis and linear mixed effects modeling (PCA-LME) were utilized to relate cartilage in vitro spectra and biomechanical properties, as well as to account for the spatial dependency. Additionally, a k-nearest neighbors (kNN) classifier was employed to reject outlying ex vivo NIR spectra resulting from a non-optimal probe-cartilage contact. Model performance was evaluated for both in vitro and ex vivo NIR spectra via Spearman's rank correlation (r) and the ratio of performance to interquartile range (RPIQ). Results: Regression models accurately predicted cartilage thickness and biomechanical properties from in vitro NIR spectra (Model: 0.77 r 0.87, 2.03 RPIQ 3.0; Validation: 0.74 r 0.84, 1.87 RPIQ 2.90). When predicting cartilage properties from ex vivo NIR spectra (0.33 r 0.57 and 1.02 RPIQ 2.14), a kNN classifier enhanced the accuracy of predictions (0.52 r 0.87 and 1.06 RPIQ 1.88). Conclusion: Arthroscopic NIRS could substantially enhance identification of damaged cartilage by enabling quantitative evaluation of cartilage biomechanical properties. The results demonstrate the capacity of NIRS in clinical applications.

Machine Learning Classification of Articular Cartilage Integrity Using Near Infrared Spectroscopy

“…In conclusion, NIRS combined with machine learning techniques, could provide a powerful tool for classification of cartilage integrity, with the potential for accurately distinguishing between normal and early osteoarthritic cartilage. This finding, combined with recent application of NIR for estimating cartilage biomechanical properties in human cadaver knee joints, 25,29 is insofar significant as it suggests that NIRS can be adapted for rapid diagnosis of cartilage integrity during knee arthroscopy, where it may be critical to correctly discriminate between healthy and degenerated cartilage prior to removal of the degenerated tissue. However, our approach needs further validation with human samples prior to clinical applications in cartilage/joint repair surgery.…”

“…To ensure sample preservation and physiological conditions, each sample's surface was covered with 0.15 M phosphate-buffered saline (PBS) solution at room temperature during spectral measurements. Although the entire probe tip (diameter = 3.2 mm) did not always cover the sample surface as a result of the curvature, the central region consisting of the receiving fibres (diameter = 1 mm) of the probe window (diameter = 2 mm) 25 was always in full contact with the sample surface during spectral acquisition. Contact was made between the probe tip and sample surface before spectral acquisition to avoid spectral absorption from the surrounding PBS.…”

Raman Spectroscopy-based Analysis of Cartilage Composition with Applications in Finite Element Modeling

“…This shows non-destructive Raman measurements made through-the-surface, similar to reported uses in arthroscopic surgeries [32,33] could be of use for FE modeling. This would have direct applications in improving modeling of OA as it develops, however the linearly scaling GAG relationship must be evaluated in degenerated cartilage samples.…”

“…However, use of vibrational spectroscopy in arthroscopic joint surgeries has recently shown potential in this area [31][32][33]. Near-infrared spectroscopy, with the use of machine learning, has demonstrated reliable predictions of proteoglycan content in the superficial zone of cartilage [31] and related mechanical properties arthroscopically [32]. The use of Raman spectroscopy has also shown similar potential arthroscopically for the differentiation of heathy vs. OA cartilage [33].…”

“…Further work to this would be based on assessing data obtained arthroscopically using spectrometers. The feasibility of using spectrometers for this purpose has recently been proven using near-infrared and Raman spectroscopes [31][32][33]. Successful distinction of healthy and damaged tissue in this scenario would offer an import means of assessing OA as it develops.…”