Raman Spectroscopy for Osteoporosis Assessment: Ex Vivo Feasibility of Molecular Fingerprinting of Bone

Raman spectroscopy is a molecular fingerprinting technique that uses inelastic light scattering to characterize material composition and has been widely applied to bone quality assessment. In this ex vivo feasibility study, we tested whether Raman spectroscopy could predict femoral neck T-scores (n = 58) from specimens obtained during elective total hip arthroplasty and classify them as normal, osteopenic, or osteoporotic. Bone samples were excited using a near-infrared, multiple spatial mode, semiconductor laser (830 nm, 150 mW), and scattered light was collected for spectral analysis. Custom MATLAB code calculated Raman outcomes. T-scores were determined from bone mineral density acquired from dual-energy X-ray absorptiometry. Raman components were correlated with T-score using simple linear or partial least squares regressions (PLSR). Bone health classifications were assessed through receiver operating characteristic curves (ROC), area under the ROC curves (AUC), and descriptive statistics. Simple linear regression models identified mineral maturity/crystallinity (MMC) as the strongest single-variable predictor of T-score in males (R² = 0.20, 95% CI 0.00-0.43) and females (R² = 0.22, 95% CI 0.00-0.46). MMC showed fair discrimination between normal and osteopenic or osteoporotic groups in males (AUC = 0.76, 95% CI 0.56-0.97; 0.77, 95% CI 0.55-0.99) and females (AUC = 0.73, 95% CI 0.52-0.95; 0.70, 95% CI 0.47-0.92). PLSR (including mineral-to-matrix ratio, carbonate-to-phosphate ratio, MMC, and pyridinoline) slightly improved combined sex T-score correlation (R² = 0.28, 95% CI 0.05-0.55) and male normal versus osteoporotic classification (AUC = 0.89, 95% CI 0.75-1.00). This ex vivo feasibility study demonstrates the potential of Raman spectroscopy to differentiate bone quality.