Two-dimensional correlation (2D) method for improving the accuracy of OCT-based noninvasive blood glucose concentration (BGC) monitoring.

The optical scattering coefficient (μs) in the dermis layer of human skin obtained with optical coherence tomography (OCT) has shown to have a strong correlation with the blood glucose concentration (BGC), which can be used for noninvasive BGC monitoring. Unfortunately, the nonhomogeneity in the skin may cause inaccuracies for the BGC analysis. In this paper, we propose a 2D correlation analysis method to identify 2D regions in the skin with μs sensitive to BGC variations and only use data in these regions to calculate μs for minimizing the inaccuracy induced by nonhomogeneity and therefore improving the accuracy of OCT-based BGC monitoring. We demonstrate the effectiveness of the 2D method with OCT data obtained with in vivo human forearm skins of nine different human subjects. In particular, we present a 3D OCT data set in a two-dimensional (2D) map of depth vs. a lateral dimension and calculate the correlation coefficient R between the μs and the BGC in each region of the 2D map with the BGC data measured with a glucose meter using finger blood. We filter out the μs data from regions with low R values and only keep the μs data with R values sufficiently high (R-filter). The filtered μs data in all the regions are then averaged to produce an average μs data. We define a term called overall relevancy (OR) to quantify the degree of correlation between the filtered/averaged μs data and the finger-blood BGC data to determine the optimal R value for such an R-filter with the highest obtained OR. We found that the optimal R for such an R-filter has an absolute value (|R|) of 0.6 or 0.65. We further show that the R-filter obtained with the 2D correlation method yields better OR between μs and the BGC than that obtained with the previously reported 1D correlation method. We believe that the method demonstrated in this paper is important for understanding the influence of BGC on μs in human skins and therefore for improving the accuracy of OCT-based noninvasive BGC monitoring, although further studies are required to validate its effectiveness.