| Near-infrared diffuse spectroscopy(NIRS)can be used to non-invasively assess the blood condition of deep tissues(up to a few centimeters).At the same time,diffuse correlation spectroscopy(DCS)can be used to extract blood flow index(BFI)from the light field time autocorrelation function).At present,this non-invasive detection method is widely used,but some non-uniform biological tissues such as brain blood information measurement,due to the influence of surface tissue,the actual measured cortical signal is noisy.In order to obtain accurate blood flow values,the research team proposed an Nth-order linear(NL)algorithm to extract blood flow values of tissues with arbitrary geometric shapes.However,for shallow interference information carried in deep signals,the removal effect of this algorithm is not obvious.In this thesis,the combination of the NL algorithm for blood flow signal calculation and the least square method is proposed and the purpose is to remove the shallow blood flow signal carried in the deep tissue blood flow.The innovative work in this article includes:Firstly,model establishment: The application of near-infrared diffused light technology has this feature: the depth of tissue measured by light source detectors at different distances is different,and the signals obtained are also different.In terms of complex non-uniform tissue head structure,for the measurement of cerebral cortex information,the optical signal measured at a longer distance is divided into two parts,the signal of the cortex and the cerebral cortex,which is beneficial to the accuracy of the measurement of the cerebral cortex signal.Therefore,the establishment of the head model needs to be roughly divided into three layers according to the actual situation,including scalp,skull,and cerebral cortex,and the corresponding optical characteristic model is established through Monte Carlo(MC)simulation for research.Secondly,algorithm research: In this thesis,based on the established head model,the organization layered design is introduced,the idea of least squares is introduced and combined with the NL algorithm,and the shallow noise signal in the deep signal is removed by data fitting,thereby obtaining a higher accuracy Target tissue blood flow.Thirdly,experimental design: The flow of the phantom solution is used in the pipeline to simulate the blood flow changes in the brain.At this time,the design of the phantom flow rate cannot be designed as the flow rate signal of different waveforms like the simulation experiment.However,during the phantom experiment,two layers of upper and lower glass tubes can be built to simulate the surface layer(scalp layer)and deep tissues(cerebral cortex).A syringe pump can be used to control the flow rate of the phantom solution to form rectangular pulse signals with different amplitudes and periods in the glass tubes.The design of the scheme provides a good platform for algorithm verification.In order to better verify the effect of the algorithm,this study was verified by computer simulation,phantom experiment and human experiment,and quantitative evaluation was carried out combining parameters such as correlation coefficient.In computer simulation,different waveforms are used to represent the two layers of tissue signals,and combined with information such as Monte Carlo simulation of photons to generate DCS data;accordingly,the phantom experiment is designed and built,thereby obtaining DCS measurement data.In addition,DCS measurements were obtained in a 70 ° head-up tilt experiment with 10 healthy volunteers.Data analysis results show that whether it is computer simulation,phantom experiment or human experiment,the combination of algorithms has a better effect on extracting deep tissue information.The research results of this thesis greatly improve the accuracy of the extraction of blood information by the near-infrared diffused light technology,and also have important guiding significance for the clinical application of this technology. |
PDF Full Text Request