Blood Flow Velocity Detection of Nailfold Microcirculation Based on Spatiotemporal Analysis

Abstract

Nailfold microcirculation can reflect the state of health. It is an important research topic to detect the change of microcirculation blood flow velocity. In order to save cost, people usually choose the low-cost micro camera to collect samples. However, the microvascular video collected by such camera is often disturbed by various noises, resulting in poor contrast and clarity of the image. Due to the large number of microvessels, it is time-consuming and laborious to detect the blood flow rate manually, which is inefficient and difficult to accurately detect the subtle changes of the blood flow velocity. At present, the previous research of blood flow velocity detection is mainly focused on the clearer microcirculation video. It is difficult to find any related research of blood flow velocity detection for noise image. For the low-quality nailfold microcirculation video collected by the low-cost microscope camera, we propose an automatic detection method of blood flow velocity based on projection analysis of spatiotemporal image. The method is as follows: firstly, video preprocessing, correlation matching are used to remove jitter and image blur is eliminated by deconvolution, the row mean is used to eliminate reflective area by analyzing the characteristics of noise distribution; secondly, we use cumulative background modeling to segment blood vessels and propose an automatic detection algorithm of blood vessel centerline; thirdly, the direction of binary spatiotemporal image is detected by using rotation projection, and then the blood flow velocity is calculated. The experimental results show that the proposed method can detect the blood flow velocity of microcirculation automatically and efficiently. Meanwhile, the average correlation coefficient between the proposed method and the manual measurement standard value is 0.935.

Blood flow characterization in nailfold capillary using optical flow-assisted two-stream network and spatial-temporal image

“…The main parameters of observation of microcirculation are vascular morphology, flow, density, curvature, etc To achieve the automation and intelligence of microcirculation diagnosis, we must first segment the blood vessel and then extract the corresponding features. Chen et al [13] and Lin et al [14] averaged the registered continuous images to obtain a vascular Mask. Wen et al [15] considered vascular detection as a supervised classification problem, using random forests to predict each pixel to segment blood vessels.…”