Numerical Optimisation of a NIRS Device for Monitoring Tissue Oxygen Saturation.

Abstract

The present work aims to develop a wearable, textile-integrated NIRS-based tissue oxygen saturation (StO2) monitor for alerting mobility-restricted individuals - such as paraplegics - of critical tissue oxygen de-saturation in the regions such as the sacrum and the ischial tuberosity; these regions are proven to be extremely susceptible to the development of pressure injuries (PI).Using a combination of numerical methods including finite element analysis, image reconstruction, stochastic gradient descent with momentum (SGDm) and genetic algorithms, a methodology was developed to define the optimal combination of wavelengths and source-detector geometry needed for measuring the StO2 in tissue up to depths of 3 cm. The sensor design was optimised to account for physiologically relevant adipose tissue thicknesses (ATT) between 1 mm and 5 mm. The approach assumes only a priori knowledge of the optical properties of each of the three tissue layers used in the model (skin, fat, muscle) based on the absorption and scattering coefficients of four chromophores (O2Hb, HHb, H2O and lipid).The results show that the selected wavelengths as well as the source-detector geometries and number of sources and detectors depend on ATT and the degree and volume of the hypoxic regions. As a result of a genetic algorithm used to combine the various optimised designs into a single sensor layout, a group of four wavelengths was chosen, coinciding with the four chromophores and agreeing very well with literature. The optimised number of source points and detector points and their geometry resulted in good reconstruction of the StO2 across a wide range of layer geometries.