Spectroscopic measurement of adipose tissue thickness and comparison with ultrasound imaging

Abstract

Near-infrared spectroscopy (NIRS) is widely applied for applications monitoring skeletal muscle oxygenation. However, this method is obstructed by the subcutaneous adipose tissue thickness (ATT) which might vary between < 1 mm to more than 12 mm. Though diffuse optical imaging can be applied to measure ATT, the objective here is to get this measure from spectroscopic data of a single source-detector distance. For the measurement of the optical lipid signal we used a broad band spatially resolved system (SRS), which is based on measurements of the wavelength dependence of the attenuation A for source detector distances p between 29 mm and 39 mm. Ultrasound images served as an anatomical reference of the lipid layer. The measurements were taken on 5 different muscle groups of 20 healthy volunteers, each for left and right limbs, e.g. vastus medialis, vastus lateralis and gastrocnemius muscle on the leg and ventral forearm muscles and biceps brachii muscle on the arm. Different analysis strategies were tested for the best calculation of ATT. There is a good non-linear correlation between optical lipid signal and ultrasound data, with an overall error in ATT prediction of about 0.5 mm. This finding is supported experimentally by additional MRI measurements as well as a multi-layer Monte Carlo (MC) model. Based on this data of the ATT thickness, a newly developed algorithm which exploits the wavelength dependence of the slope in attenuation with respect to source-detector distance and MC simulation for these parameters as a function of absorption and scattering coefficients delivers a considerably better fit of reflectance spectra when fitting haemoglobin concentrations. Implications for the monitoring of muscle oxygen saturation are discussed.