Research Of ID-OCTA For Blood Flow Imaging And Hemodynamic Quantification

Optical coherence tomography(OCT)is an important biomedical imaging technology.By measuring endogenous blood flow motion contrast,OCT angiography(OCTA)developed based on OCT can achieve non-invasive,micron-level resolution,depth-resolved three-dimensional real-time imaging of blood flow in biological tissues,which is widely used in ophthalmology,brain science and other fields.Among different OCTA methods,the complex decorrelation based OCTA extracts flow signals by calculating the correlation between signals from different time points,and has the potential for the quantification of hemodynamic parameters at the same time.This thesis studies the application of this decorrelation-based OCTA(represented by ID-OCTA)in the fields of blood flow imaging and hemodynamic quantification,including:First,due to the significant dependence between decorrelation and signal-to-noise ratio(SNR)for OCT signals,it is difficult to accurately distinguish dynamic blood flow signals from low-SNR static tissue signals based on decorrelation alone.To solve this problem,a blood flow imaging method based on inverse SNR(iSNR)and decorrelation features,which is named as ID-OCTA(inverse SNR and decorrelation based OCTA),is proposed.By deriving the asymptotic relationship between decorrelation and iSNR theoretically,it is proved that in the decorrelation-iSNR feature space,a straight line can be used to accurately distinguish the blood flow signals from the static tissue signals.And the boundary line can be determined using a method based on numerical simulation.The reliability of the simulation method is verified by the in vitro phantom experiment,and the accuracy improvement of flow extraction is quantitatively proved.Besides,it is proved on the human cheek skin imaging data that ID-OCTA can obtain a higher contrast blood flow image compared with the traditional intensity mask method.For the problem of tail artifacts in OCTA blood flow images,a new artifacts removal method is proposed.The real blood vessels and tail artifacts are distinguished based on the gray scale of ID-OCTA(the product of signal intensity and decorrelation),and the artifacts are filtered depth by depth.In the imaging experiment of the three-layer vessel structure in mouse retina,it is verified that this method can effectively remove the artifacts and at the same time restore the deep vessels blocked by artifacts.Furthermore,in terms of hemodynamic quantification,the dynamic range and uncertainty of the current decorrelation-based methods are limited by the ensemble size.To solve this problem,an adaptive spatial-temporal kernel for decorrelation calculation is proposed,which enlarges the ensemble size by collecting samples in both space and time dimensions.And the impact of interference in the time dimension is reduced with the help of the adaptive size design.Numerical simulation experiments and phantom experiments are carried out to verify the effect of spatial-temporal kernel to expand the dynamic range and suppress the uncertainty.Meanwhile,in the rat cortex electrical stimulation experiment,the proposed method is used to monitor the hemodynamic response,and the adaptive design is proved to be capable of suppressing the disturbance.It is also verified that the proposed method can improve the resolution of different response curves,showing a promising application prospect in hemodynamic research.