Fluorescence spectroscopy precorrection model robust to large variability of optical properties: application to point of care quantitative liver graft viability assessment.

Optical properties correction models for fluorescence spectroscopy faced limitations due to the high variability of optical properties. Typically, the point-of-care assessment of liver graft viability remains a critical challenge in transplantation medicine, where current methods rely on subjective visual inspection and delayed laboratory biomarkers. While optical approaches offer promising solutions to this challenge, the unique characteristics of liver grafts, which undergo highly varying conditions from procurement to transplantation, ranging from blood perfusion to ischemic phases and perfusion with transparent solutions, create an extremely wide dynamic range of optical properties that challenge state-of-the-art optical models. We present a novel method combining fluorescence spectroscopy and diffuse reflectance measurements, accounting for the large variability in optical properties and adapted for liver graft viability assessment. Our approach corrects for tissue optical properties using a dual-modality system that measures both fluorescence and diffuse reflectance spectra, enabling accurate quantification of metabolic fluorophores such as NADH and FAD. The method was validated on tissue-mimicking phantoms with varying optical properties and demonstrated superior performance compared to state-of-the-art correction techniques, with quantification accuracy improved by at least a factor of 1.5 and a capacity to detect more than 92% of the fluorophores present in the different phantoms. Clinical validation was performed on porcine ischemia-reperfusion models, showing promising results for real-time tissue viability monitoring during organ preservation and transplantation procedures.