OBJECTIVEThe assessment result of microcirculatory perfusion provides significant evidence for the detection of microcirculatory alterations in sepsis patients, and functional capillary density based on vessel segmentation can reflect microcirculatory perfusion effectively.

METHODSFirst, preprocessing comprise operations to improve the quality of images in order to maximize the contrast between background and vessels. Secondly, adaptive local thresholding is used for initial segmentation to get the general vessel region. Finally, post-processing technique reduces the image artifacts and links the vessel fragment, and then functional capillary density is calculated based on the segmentation results.

RESULTSSeveral sublingual microcirculatory data have been tested, and the agreement between the results by AVA and the proposed algorithm reached up to 92%.

CONCLUSIONSThe method can realize automated assessment of sublingual microcirculatory perfusion rapidly and accurately, and has strong clinical value.