Estimation of deoxygenated and oxygenated hemoglobin by multispectral blind linear unmixing

Blood perfusion parameters can be used to evaluate the micro-circulatory health condition of a patient. Several non-invasive optical techniques have been used to estimate blood perfusion as near-infrared spectroscopy or pulse-oximetry. However, these techniques require contact with the patient, and the measurements are restricted to a single point evaluation. These disadvantages could be solved by multispectral imaging. Hence, this paper presents an approach based on multispectral imaging and blind linear unmixing, as an alternative to estimate blood perfusion parameters in the hand palm. This work evaluated changes in oxygenated and deoxygenated hemoglobin concentrations by employing an experimental occlusion protocol in healthy volunteers. We compared the results of several blind linear unmixing and linear regression models. The average cosine similarity values between the prediction model and the photoplethysmography estimations varied in the range 87% and 96%. The mean R-squared adjusted values for oxygenated and deoxygenated hemoglobin were greater or equal than 0.75 and 0.84, respectively. Our results demonstrated the feasibility of non-invasive estimation of hemoglobin in the hand palm, and opened the possibility for calculating other perfusion parameters that help to diagnose and monitor pathologies in large tissue regions.

Automated summary

This study presents an approach based on multispectral imaging and blind linear unmixing to estimate blood perfusion parameters in the hand palm. The results demonstrate the feasibility of non-invasive estimation of hemoglobin and the potential for calculating other perfusion parameters to diagnose and monitor pathologies.