Optimal determination of detector placement in cerebral NIR spectroscopy of neonates using chemometric techniques

Abstract

This paper investigates the optimal placement of NIRS optodes in order to maximise the detection of haemoglobin changes in cortical grey matter resulting from an evoked response in neonates. The analysis is based upon predictions of optical signal at the surface of the head, using a Finite Element based model of light diffusion in tissue. Using the generated intensity data, the combination of optode positions, which maximise the signal from cortical grey matter whilst minimising that from surface tissue or cerebral white matter, is determined using a Chemometric statistical analysis. The neonatal head is modelled as a 2 dimensional circle with 3 layers corresponding to the skin/scalp, and grey and white matter. A wide range of absorption coefficients for each layer is simulated, based upon physiologically reasonable values for parameters. Surface intensity at 10 different optode positions have been generated for a total of 31,250 combinations of these variables for the 3 layers. It was found that with 3 optodes at 5, 15, and 50 mm apart from the source, the smallest root-mean-square error between the estimated and modelled values can be obtained. Increasing the number of optodes further does not improve the performance.