Microfluidic phenotyping of cilia-driven mixing for the assessment of respiratory diseases

The function of ciliated surfaces to clear mucus from the respiratory system is important for many respiratory diseases and, therefore, has a high impact on public health. In this work, we present a quantitative method to evaluate mixing efficiency of cilia-driven microfluidic flow based on front line deformation as an integrated measurement of cilia function. So far, mixing efficiency has been used mainly for analyzing artificial cilia. Most of this work, however, was either bound to specific imaging modalities or done on simulated data. In this simulations, mixing efficiency has been quantified as the change in length of a virtual dye-strip. We adopt this measure for in-vivo data of the Xenopus tropicalis tadpole that is acquired by an innovative low-cost mixing assay (microscopy) and optical coherence tomography (OCT). Mixing is imaged in a water filled well while dye flows into it. The length of front line is extract with the following steps: (i) filtering of the video to reduce compression artifacts, (ii) segmentation of dye based on the hue channel in HSV colorspace, (iii) extracting and converting the front line of segmented dye to curvature scale space, and (iv) smoothing of the front line with a Gaussian filter and calculation of length in curvature scale space. Since dye cannot be used with OCT, we use data from prior work that performs particle tracking to generate a flow vector field and seed virtual dye in this flow field. The following steps extract the vector field: (i) filtering and gray scale thresholding for particle candidate detection, (ii) thresholding size of particle candidates, (iii) pairing of remaining particles from subsequent frames, (iv) estimation of velocity and direction of each particle, and (v) combining these measures into a velocity field. Our in-vivo imaging and analysis shows that the front line of dye is actively mixed by the ciliated surface of the Xenopus embryo.