Spectroscopic roadmap to non-ionizing radiation bone screening

We present a non-invasive, label-free approach for assessing bone quality using transcutaneous spatially offset Raman spectroscopy (SORS) combined with PLS2 (multi-response partial least squares) regression. A custom fiber-optic probe was used to collect Raman spectra from human cadaver fingers at multiple lateral source–detector offsets (0, 3, and 6- mm), a geometry optimized to enhance sensitivity to subsurface bone tissue. To validate this approach, Raman spectra were acquired from 21 human cadaver donors spanning all three clinical bone health categories: normal, osteopenia, and osteoporosis. For each donor, spectra were collected both transcutaneously and directly from exposed bone at anatomically matched sites. Using paired transcutaneous and exposed-bone datasets, we developed a supervised multivariate PLS model to estimate bone-specific Raman spectra from transcutaneous measurements. The reconstructed spectra faithfully reproduced key biochemical features of bone and enabled extraction of established Raman-derived metrics, including mineral-to-matrix band area ratios and full width at half maximum (FWHM) parameters known to correlate with bone composition and mechanical integrity. Across donors, metrics such as PO₄³⁻/CH2, PO₄³⁻/Amide I, ratios and inverse CH₂ FWHM (1/FWHMCH2) preserved the expected diagnostic ordering and demonstrated statistically significant separation between normal and osteoporotic groups. These findings demonstrate that transcutaneous SORS, when combined with regression-based bone signal recovery, can non-invasively capture Raman biomarkers indicative of bone quality, supporting its potential as a clinical tool for osteoporosis assessment and monitoring.