Improvement On Measuring Accuracy Of Flow Speed For Laser Speckle Contrast Imaging

Laser speckle contrast imaging (LSCI) has been widely used in detecting blood flow changes in brain, skin, retina, arthrosis and mesentery with the advantages of fast processing, high spatial and temporal resolution, full-field imaging without scanning, providing important tool in investigating functional activities of tissues, exploring mechanisms of diseases and evaluating drug efficiency. Meanwhile, LSCI has significant potential application in detecting seed vigour with the advantages of fast imaging, non-contact and simple system structure, thus provides potential tool for seed breeding and crop store. However, many problems deserve addressed, improved and investigated intensively, either in the methodology and the application aspects of LSCI.Therefore, this thesis investigates problems in LSCI in both the methodology and the application aspects. In the methodology aspect, we investigate some important problems in LSCI through theoretical arithmetic, numerical simulation and phantom experiments, including:the relation between spatial resolution of LSCI and statistical accuracy; impact of noises in imaging system on statistical accuracy of speckle contrast; linear response range of LSCI in speed estimation; impact of speckle size on static and dynamic speckle contrast values; comparisons among the existing speckle contrast analysis methods. In the application aspect, we verify the vaidity of LSCI in seed vigour test, and investigate the performances of some laser speckle image processing methods and the optimum imaging conditions for these methods. The main results of this thesis include:(1) To overcome limitations of the existing laser speckle simulation methods, we develop a time-integrated dynamic image speckle simulation method based on copula. Combining this method with phantom experiments, we obtain the following results:We point out that in the precondition of maintaining statistical accuracy, spatial resolution of LSCI can only be as high as the half of the maximum spatial resolution of the optical imaging system; Shot noise and DC noise have significant impacts on the statistical accuracy of speckle contrast under weak light illumination. The relative error in speckle contrast can be higher than 10% when the mean intensity is lower than 200 counts. Differences in Brownian motion lead to linear response range of LSCI for different scattering samples.(2) We investigate impacts of speckle size on LSCI. We present an approximative relation among local speckle contrast of static speckle, speckle size and length of spatial window, providing guidance for selecting the optimum speckle size in LSCI. Investigations on dynamic speckle show that spatial integration effect of CCD pixels has equal impact on spatial and temporal contrast. However, speckle size significantly affects effective statistical numbers in spatial domain, but has little impact on effective statistical numbers in temporal domain, which explains why temporal contrast analysis requires fewer statistical numbers than spatial contrast analysis does to achieve accurate speckle contrast.(3) We compare the existing speckle contrast analysis methods whih are based on combination of spatial and temporal statistics. Both simulation and animal experiment results show that the maximum relative error in contrast for spatio-temporal speckle contrast analysis method (stLASCA) is 5%, but such measure can be higher than 13% for both temporal averaged spatial speckle contrast analysis method (sLASCA) and spatial averaged temporal speckle contrast analysis mehotd (tLASCA). The statistical noises of these three methods are in the same level. The above results suggest that stLASCA achieves higher statistical accuracy than sLASCA and tLASCA do.(4) Image processing results for the speckle images of soaked corn seeds show that LSCI (specifically sLASCA and tLASCA) and laser speckle time-difference imaging method achieve fast seed vigour detection within 10 s, with image contast of 0.26,0.32 and 0.57, respectively. Image processing results for the speckle images of corn seed in early germination state show that the laser speckle time-difference imaging method achieves the highest image contrast when setting exposure time to 5 ms, but it is unable to differentiate the detailed differences in speeds. On the contrary, difference in flow speeds between hypocotyls and other parts of the embryo can be discriminated using LSCI when setting exposure time to 500 ms and higher.