Image Processing Algorithm to Extract White Blood Cell Counts From Non-invasive, In-vivo, Time-lapse Images of Nailfold Capillaries

Systems, apparatus, and methods for non-invasively detecting a white blood cell (WBC) event from in vivo capillary data may include obtaining a plurality of images, followed by specifying two-dimensional (2D) coordinates corresponding to internal contour and external contour of a capillary visible in the images. The two sets of 2D coordinates may be resampled to generate two sets of resampled coordinates, between which intermediate curves, including a middle curve, are generated. Curvilinear distances may be defined based on the middle curve. Intensity values, each of which corresponds to one of the images, one of the intermediate curves, and one of the curvilinear distances, may be extracted and transformed to the Radon domain. A plurality of maxima locations in the Radon domain corresponding to a flow trajectory inside the capillary may be identified. Detection of a visual gap in the flow trajectory inside the capillary indicates a WBC event.