Image processing and machine learning for diagnostic analysis of microcirculation

摘要

Automated quantitative analysis of microcirculation, such as density of blood vessels and red blood cell velocity, is implemented using image processing and machine learning techniques. Detection and quantification of the microvasculature is determined from images obtained through intravital microscopy. The results of quantitatively monitoring and assessing the changes that occur in microcirculation during resuscitation period assist physicians in making diagnostically and therapeutically important decisions such as determination of the degree of illness as well as the effectiveness of the resuscitation process. Advanced digital image processing methods are applied to provide quantitative assessment of video signals for detection and characterization of the microvasculature (capillaries, venules, and arterioles). The microvasculature is segmented, the presence and velocity of Red Blood Cells (RBCs) is estimated, and the distribution of blood flow in capillaries is identified for a variety of normal and abnormal cases.

主权项

1. A computer implemented process for the diagnostic analysis of microcirculation, comprising the steps of:
obtaining video images of microvasculature from tissue surface videos; stabilizing by computer the obtained video images to eliminate motion artifacts; segmenting by computer blood vessels from each video frame of stabilized video images to provide quantitative and qualitative measure of microcirculation; identifying by computer active capillaries in stabilized and segmented images and labeling blood vessels as to quality of flow; calculating by computer Functional Capillary Density (FCD) and Proportion of Perfused Vessels (PPV) in identified active capillaries; and based on calculated FCD and PPV, classifying by computer subjects as normal or hemorrhagic.