Real-time Analysis of Microvascular Blood Flow for Critical Care

Microcirculatory monitoring plays an important role in diagnosis and treatment of critical care patients. The Sidestream Dark Field (SDF) imaging devices have been developed to visualize and support interpretation of the micro-vascular blood flow. However, due to subsurface scattering within the tissue that embeds the capillaries, transparency of plasma, imaging noise and lack of focus, it is difficult to obtain reliable physiological data from SDF videos. Therefore, thus far microcirculatory videos have been analyzed manually with significant input from expert clinician. In this paper, we present a framework that automates and reduces subjectivity of the process. It includes stages of video stabilization, enhancement, and micro-vessel extraction, in order to automatically estimate statistics of the micro blood flows from SDF videos. Our method has been validated in critical care experiments conducted carefully to record the microcirculatory blood flow in test animal subjects before, during and after induced bleeding episodes, as well as to study the effect of fluid resuscitation. Our method is able to extract microcirculatory measurements that are consistent with clinical intuition and it has a potential to become a useful tool in critical care medicine.


" Real-time Visual Analysis of Microvascular Blood Flow for Critical Care "
Chao Liu, Hernando Gomez, Srinivasa G. Narasimhan, Artur Dubrawski Michael R. Pinsky and Brian Zuckerbraun
IEEE Conference on Computer Vision and Pattern Recognition (CVPR) 2015.
[PDF] [Poster]


Sidestream Dark Field Imaging. (a) Portable SDF imaging device used for microcirculatory monitoring. (b) One frame of the microcirculatory video. (c) The LEDs, arranged and optically isolated around the lens system, emit light optimized for red blood cell absorption.
The vessel skeletons are extracted from the minimal image across the frames. First column: the first frames of the videos. Due to subsurface scattering and transparency of plasma, it is hard to detect capillaries from a single frame. Second column: denoised minimal image across all the N frames. In our case N = 200. Third column: extracted vessel skeletons. Fourth column: manually painted vessels. The index and status of subjects in each row: Pig 50, before resuscitation; Pig 53, end of baseline; Pig 60, end of baseline.
The first frame in the original video and the motion-magnified video.
The averaged observed motions (blue) across the frame and their components. Motion components due to heartbeat and breathing are colored in red and green respectively.
Blood flow velocity estimation. (a) The extracted vessel skeletons. The vessel segment for which the flow velocity is estimated is colored in blue. (b) The EPI image of the blue colored vessel segment in (a). (c) The Fourier Transform of the EPI image. (d) The dominant orientation of the Fourier Transform is plotted as the green line. The corresponding line showing the dominant orientation in the EPI image is plotted in red in (b).
Setup of the experimental procedure. 18 pigs are observed carefully at various stages of bleeding and resuscitation.
The blood flow velocity distributions at three key stages of pigs


Because of imaging noise, relative motion between sensor and subjects, low contrast due to sub-surface scattering, and lack of texture, conventional methods heavily relied on the appearance features will not work for the micro-vascular video.

Video stabilization process. The heartbeat and breathing are also estimated during this process.

The vessel skeletons are extracted from the minimal image across the frames.

Blood flow velocity estimation.

The blood flow velocity distributions at three key stages of pigs.


This research was supported in parts by NSF awards 0964562 and 1320347, NIH grants 1K12HL109068-02 and 1R01GM082830-01A2, a Veterans Affairs Merit Award 1I01BX000566 and a Department of Defense grant DM102439.