Study Of Fourier Domain Optical Coherence Tomography Based 3D Blood Flow Imaging Of Human Skin

Optical coherence tomography(OCT)was proposed in the 1990s and is a promising biomedical imaging technique with the advantages of high resolution,high imaging velocity and non-invasion.In OCT,the sample information at different depths is extracted with the interference fringes of reference light and back-scattered light from tissue,then the 3D structural image of sample can be obtained by scanning of probe beam in two transverse directions with two scanners.The resolution of OCT method is at micron level and the maximum imaging depth in tissue is about 2 mm.Nutrients and oxygen are transported to entire body through blood and the material exchange occurs in capillary vessels,so the distribution of blood vessels will undergo corresponding changes when lesions occur in tissue.Microcirculation maps of tissue can then be helpful for diseases diagnosis,such as cancer,psoriasis and port wine stain.In this thesis,a spectral domain OCT(SDOCT)was first designed and built,the performance of which was then analyzed and optimized.The thesis is mainly foucused on OCT based blood flow imaging.The main work includes the following:An SDOCT system was designed and built for imaging tissue,a rapid scanning optical delay line was then built to compensate the dispersion difference between reference arm and sample arm.To improve the precision of dispersion compensation,the value of dispersion difference was calculated to evaluate the performance of dispersion compensation.A new method based on this system was further proposed to measure the dispersion and refractive index of transparent sample.Cross-correlation in spectral domain based phase-resolved Doppler OCT was proposed to extract blood flow signals,in which two adjacent A-scans of interference fringes were correlated to suppress the random phase noises caused by unstable system and environment vibration.The signal to noise ratio of Doppler frequency shift was improved from 53 dB to 56 dB in phantom experiments and from 45 dB to 49 dB in in vivo experiments.Cube data based correlation mapping OCT was proposed to reconstruct the microcirculation maps in tissues.The signal to noise ratio of blood flow signals can be improved by this method without decreasing the resolution in the XZ cross section.The experimental results demonstrated that the signal to noise ratio of blood flow signals extracted by cube data based correlation mapping OCT with window size of 3x3x3 pixels was improved by 2 dB and the resolution of blood flow image was doubled compared to that obtained by conventional correlation mapping OCT with window size of 7x7 pixels.Zero-padding technique and cross-correlation based image registration technique were presented to remove the noise caused by motion artifacts due to cardiac and respiratory motions.At the same time,the imaginary part based structural images were used to calculate correlation value to improve the sensitivity for blood flow detection.The results demonstrated that the motion artifact noise was removed by the zero-padding technique and image registration technique in human skin experiment and some more small blood vessels were detected with imaginary part based method in mouse ear experiments.Multi-beam scanning method was proposed to suppress the noise in blood flow images,and the influence on Doppler phase and speckle variance was investigated when the sample was located out of the beam focus.The results of human skin experiments demonstrated that the quality of the blood flow images obtained by phase resolved Doppler OCT,phase resolved Doppler variance OCT and speckle variance OCT methods was improved by the multi-beam scanning method.