High Resolution Laser Speckle Imaging Of Blood Flow: Methods And Applications

Based on the coherent scattering properties of moving particles, laser speckle contrast imaging (LSCI) provides relative velocity information. Using this technique, we can esti-mate two dimensional cerebral blood flow. LSCI is a powerful tool in the investigations of tissue perfusion and cerebral blood flow under physiological and pathological conditions. We use CCD or CMOS camera to record the images of illuminated area and then calculate the contrast image by laser speckle contrast analysis (LASCA). The contrast image is related to the velocity distribution. Based on the relation between velocity and contrast, we can further estimate the relative velocity of moving particles, i.e. blood cells.Although traditional LSCI provides the contrast image with a good spatiotemporal res-olution, problems are still existing:1) because of the imaging noise and ambient effect in the background, the dynamic range of contrast data is much limited and thus make it difficult to visualize and analyze the data; 2) the inhomogeneous effects due to non-uniform distribution of laser intensity and the curvature of imaging plane may bias the contrast estimation; 3) dis-turbances due to the breath, heart beating and other noises decrease the spatial resolution; 4) spatio-temporal non-stationarity in velocity of CBF results in a higher noise level in contrast analysis and thus lower SNR in contrast data. All these problems have to be properly solved in order to get high SNR and high spatio-temporal resolution of blood flow imaging. Further more, the traditional LSCI device has big size and complex part connections which results in difficulties when applying to different research labs. The animal's cerebral blood flow can not be monitored in awake or freely moving status either.The aim of this thesis is to develop the techniques and new devices to solve the above problems respectively, i.e. dynamic range enhancement of contrast data based on monotonic point transformation; model-based correction for inhomogeneous effect; spatial resolution enhancement by registration technique; high SNR and high spatio-temporal imaging using random process estimator. Based on these techniques, we investigate the angiogenesis after mice ear injury. We also develop the new integrated imaging system. With the new imaging system, we investigate the cerebral blood flow changes in mice model of middle cerebral artery occlusion (MCAO). We also design a miniature laser speckle imager for full-field high-resolution imaging of cerebral blood flow in freely moving rats. The main contents and results are summarized as follows:(1) To enhance the limited dynamic range of contrast data, the enhanced laser speckle contrast analysis (eLASCA) based on monotonic point transformation is proposed. eLASCA can enhance the dynamic range efficiently while keeping the data precision. Furthermore, the relation between eLASCA data and velocity is obtained. In practice, eLASCA also improves the segmentation of blood vessels, which can also be used in automatic visualization of both contrast image and velocity image. Using eLASCA, the tissue perfusion of rat's cerebral cortex under hypothermia and re-warming conditions are investigated. The results show that hypothermia leads a lower perfusion level than normothermia (ratio of contrast values is 189%). Based on paired t-test, the velocity increase is significant from hypothermia to re-warming conditions, meanwhile eLASCA method provides a higher confidence level: eLASCA (p=0.009), LASCA(p=0.013).(2) A model-based correction method is proposed to eliminate the inhomogeneous ef-fects due to non-uniform distribution of laser beam and the curvature of imaging plane. A mathematical model of inhomogeneous effects is constructed and then estimated by nonlin-ear curve fitting method to reconstruct the contrast image. In an application of simultaneous imaging of blood flow and de-oxygen hemoglobin of cerebral cortex, this model-based cor-rection is successfully implemented to obtain the accurate contrast and hemoglobin images, which is further applied to the segmentation and identification of cerebral arteries and veins.(3) In order to eliminate the disturbances from breath and heart beating, we proposed a registered laser speckle contrast analysis (rLASCA) method, including a convolution kernel, normalized correlation metric, cubic B-spline interpolator and then a traditional temporal LASCA. rLASCA significantly improves the spatial resolution of contrast image. We apply rLASCA to the study of angiogenesis in brain tumor. The results show obvious angiogenesis in rat's cerebral cortex 10 days after the tumor cells injection. In particular, some small new vessels can only be observed in the contrast image produced by rLASCA.(4) In order to improve the SNR under spatio-temporal non-stationarity condition, a random process estimator method is proposed. Applying random process theory, the noise models and statistical properties in spatial and temporal LASCA are developed and analyzed. Based on the statistical properties of noise models, the random process estimator is proposed to obtain high SNR and high spatio-temporal resolution contrast image. The random process estimator is then applied to to the functional electric stimulation induced spatio-temporal response of cerebral blood flow in somatosensory area of rat's cortex. For single stimulation trial, random process estimator gives a much lower estimation error (0.31±0.03) than tradi-tional temporal LASCA method (1.36±0.09). The average cortical spatiotemporal response to ten trials of stimulation showed clear activations in some specific small vessels, which was not observable by traditional temporal LSACA due to its low spatio-temporal resolution(5) Combining the above techniques, we investigated the angiogenesis after mice ear injury. In the angiogenesis study of 7 mice, it only shows part perfusion recovery 3 days after injury (contrast values are 0.5516±0.0860 on day 0 after injury and increase to 0.8481±0.0888 on day 3, with paired t-test p=0.00054<0.01), which is more significant after 6 days (contrast values 0.1263±0.0839, paired t-test with day 0 after injury p=0.00005<0.01). While the perfusion slightly decreases (contrast values 0.2992±0.0093) after 12 days compared with that on day 6 (p=0.0041<0.01), which indicates the perfusion recovery reaches a stable level. A new laser speckle imaging system with compact hardware implementation is developed. With this integrated system, we inves-tigated the cerebral blood change in mice MCAO model. The new imaging system was able to monitor the whole ischemic and re-perfusion in real time, furthermore, we were able to observe the different neurovascular response at different cortical areas during the ischemic and re-perfusion procedure, even the dynamic changes in ischemic core and penumbra can be visualized.(6) We designed a miniature laser speckle imager that weighs approximately 20 g and is 3.1 cm high for full-field high-resolution imaging of cerebral blood flow (CBF) in freely moving animals. Coherent laser light illuminates the cortex through a multi-mode optical fiber bundle fixed onto the supporting frame of the imager. The reflected lights are then collected by a miniature macro-lens system and imaged by a high-resolution CMOS camera at a high frame rate (50 fps). The proposed miniature imager can provide stable and reliable CBF imaging in freely moving animals. Using this miniature imager, we achieve high spatio-temporal resolution laser speckle contrast imaging of CBF in freely moving animals in real time.In this thesis, high SNR, high spatio-temporal resolution imaging methods and systems (compact and miniature) are developed. The new imaging methods and systems show great potentials in the investigations of brain function and brain injury.