OCT Based Studies Of Skin Features And Their Applications In The Non-invasive Blood Glucose Test

Optical Coherence Tomography can non-invasively image skin tissue with highresolution to detect and analyze the internal structure, and obtain optical parametersskin tissue to analyze the structural changes within the organ or detect varaiations ofblood glucose levels. In this paper, for expanding application of OCT technology inskin tissue, we focus on developping image enhancement algorithms to eliminate theinfluence of speckle noise in noinvasive glucose detection and attenuation on OCTimage SNR and image contrast for improving visualization of OCT imagesvisualization, researching on optical parameters of skin tissue to quantitativelyanalyze subtle structural features inside skin tissue induced by location device innoninvasive glucose detection, and studying non-invasive blood glucose detectionbased on several more sensitive areas inside skin tissue.Int this thesis, we propose an adaptive Volume-BM3D denoising algorithm bycombining non-local filtering and wavelet transform based on genearal strongsimilarity inside the three-dimensional OCT data. In the human fingertip experiment itimproves with18.4dB in SNR relative to the the original image, and effectivelydecrease the influence of speckle noise on OCT signal of noninvasive glucosedetection. Also, we strudy an adaptive multiscale Retinex attenuation compensationalgorithm by analyzing differences in principles between OCT signal attenuationcompensation and image illumination compensation. It is verified by human fingertipimage, skin tissue image and tooth image, achieve with average increase of5762inVCM relative to the the original image, and dig out the hidden structure inside tissuein the OCT original image.We analyze the principle of OCT signal, and propose a slope OCT enhancementalgorithm to enhance subtle internal structures of skin tissue, and testify theeffectiveness of the algorithm with multi-layer tape, skin tissue and tooth.Furthermore, we propose a Gamma distribution model of dimensions of interal finestructures inside tissue to quantitatively evaluate the tissue. In contrast experiments ofGamma distribution of papillary dermis layer and reticular layer of skin tissue andnormal skin tissue with scar tissue, we find that this model can quantitatively evaluates the dimensional distribution of the internal fine structure of skin tissue andits variance.The impact of pressure and osmotic media used in non-invasive blood glucosemonitoring are analyed by adaptive multiscale Retinex attenuation compensationalgorithm and Gamma distribution model, and then we propose an equipment andmethod to ensure acquiring fixed skin tissue during the experiment. In addition,correlations at different depth and range between blood glucose values and scatteringcoefficient are estabilished using the sliding window method and draw in thecorrelation figure. According to this correlation figure and selecting criteria of theglucose sensitive region, we extract multiple skin tissue glucose sensitive regions andestablish multivariate model for non-invasive blood glucose detect model based onpartial least squares. And OCT technology for non-invasive blood glucose testing hasbeen verified through four experiments that94%of blood glucose reading ofmultivariate model of OCT glucose detection fall in the A region in the Clark errorgrid analysis, which have positive effect on directing clinical diagnose.