P015 - Feasibility of Near-Infrared Reflective Spectroscopy to Measure Local Variations in Cartilage Quality: A Human and Animal Cadaveric Study
Abstract
Purpose
There is a clinical demand for a tool for peri-surgical, non-destructive evaluation of local cartilage quality. Such a tool can be used for evaluation of treatment necessity, determination of (viable) borders of a cartilage defect, and evaluation of repair tissue during second-look surgeries. Therefore, the aim of this study was to assess the feasibility of near-infrared spectroscopy as a tool for non-destructive, minimally-invasive measurement of cartilage quality.
Methods and Materials
A custom-developed 2-mm optical-fibre probe was attached to a 350-1830nm wavelength spectrometer. 450 diffuse reflectance spectra were obtained from 90 cartilage sites on eight fresh-frozen human cadaveric knees. Local glycosaminoglycan, hydroxyproline and water content; cartilage thickness; Young’s modulus (using micro-indentation); T1rho and T2-mapping MRI images; and macroscopic osteoarthritis score were obtained. With 35 spectral features, related to water, lipids and blood, a linear regression model was used to assess the relation to physiological outcome parameters. Moreover, a linear discriminant classification model was used to study the discriminative power between weight-bearing and non-weight-bearing areas. Additionally, in a mini-pig cartilage-defect model, the difference between defect and nearby control areas (40 measurements over 8 defects) was classified directly post-mortem.
Results
The linear models showed the best correlation with osteoarthritis score (R2=0.72), cartilage thickness (R2=0.70), and hydroxyproline content (R2=0.61), and the weakest with Young’s modulus (R2=0.42). Higher correlation to T2 mapping than T1rho (R2=0.55 vs 0.45) was found, suggesting higher sensitivity to matrix organization than concentration. Weight-bearing and non-weight-bearing parts were classified with high accuracy (90.9%). Spectral measurements in- and outside the cartilage defects in the mini-pigs (Fig. 1) could be classified with 100% accuracy.
Conclusion
Good correlations between spectral features and osteoarthritis-related measures were found. Further validation in vivo is needed. The discriminative classification between healthy and damaged cartilage was excellent, suggesting that applications for cartilage defect (repair) evaluation are feasible.