Blood Flow Velocity Detection Of Nailfold Microvasculature And Application

Human microvasculature usually refers to the capillary at the junction of arteriole and venule.Microcirculation refers to the blood flow in human microvasculature and its main function is to complete the material exchange required by organs and tissues of human body.Microcirculation can reflect the health of human body.The blood flow velocity of microvasculature is an important flow state parameter,which can dynamically reflect the state of human blood flow.Therefore,accurate detection of blood flow velocity is of great significance in research,and high-precision detection of blood flow velocity is a great challenge.The artificial measurement of blood flow velocity in microvascular is inefficient and it is difficult to accurately obtain the subtle changes of multiple vessels;the application of high-end medical equipment to detect blood flow velocity is relatively limited;in order to reduce the cost,people began to research the blood flow velocity detection based on video analysis.Up to now,to the utmost of our knowledge,there are few researches on automatic detection and most of these methods are used to process clearer microvascular images.However,the microvascular video samples collected by ordinary micro camera are easily interfered by noises.There are still many difficulties in automatic detection of blood flow velocity by using low-quality images.In view of the noise interference micro vascular video collected by ordinary micro camera,we propose an automatic detection algorithm of blood flow velocity based on projection analysis.Firstly,microvascular image enhancement.The microvascular video is de jittered and de blurred,and the image contrast is improved by frequency domain processing;then,the microvascular image segmentation.The whole microvascular was segmented accurately in the complex background.finally,two methods of blood flow velocity detection are proposed.The specific research of this thesis includes the following parts:(1)Image enhancement and preprocessing of nailfold microvascular.Because of the inevitable factors such as the human breath,pulse beat and the stability of the acquisition equipment.Combining the jitter amplitude and direction of the video,we proposed a video de jitter algorithm based on correlation calculation.Due to the relative motion between the camera and the sample,the image is blurred.We use deconvolution to eliminate image blur.Wavelet transform is used to improve the image and the G channel average illumination is proposed.(2)Image segmentation of nailfold microvascular.Firstly,the mean image of microvessels is obtained by accumulating G-channel images,and the line gray average is used as a threshold to eliminate the reflective area;secondly,the local window mean binarization is used to segment the enhanced G-channel cumulative mean image.It can obtain the foreground target in the microvascular image;finally,the contour information is used to customize the morphological structure elements to eliminate the noise and the cross-type blood vessels and abnormal blood vessels with short effective lengths are removed by using regional filtering.(3)Blood flow velocity detection of nailfold microvascular.There are some specific moving targets in the microvascular.The blood flow velocity can be detected by tracking the target.According to the characteristics of blood flow in microvasculature,blood flow velocity detection based on projection analysis is proposed,and the speed and direction of blood flow are calculated according to the center of gravity of differential projection histogram.On the basis of histogram projection,blood flow velocity detection based on binary spatiotemporal(ST)image is proposed.Firstly,the relationship between time and space of blood flow is analyzed,and the binary ST image of a single vessel is generated;secondly,the direction of the ST image is calculated by rotational projection;finally,the blood flow velocity is calculated.At present,there are few researches on automatic detection of blood flow velocity.There is no standard database and value,so many experiments cannot provide comparative data.Comparing with the results of manual annotation is a common way of experimental comparison.Several samples are tested and compared with the results of manual annotation in this thesis.The results show that the proposed method has achieved good results in image enhancement and vascular segmentation.The accuracy of blood flow detection is relatively high.There are two main innovations:(1)A better microcirculation image preprocessing and blood vessel segmentation method was proposed for the nail fold microcirculation video collected by low cost micro camera.(2)An effective method to detect the blood flow velocity of nail fold microvasculature was proposed,which can provide a new research idea for the further calculation of microcirculation.