Recovering the appearance of the capillary blood column from under-sampled flow data

The regular spacing of cells in capillary flow results in spurious cell trajectories if the sampling rate is too low. This makes it difficult to identify cells, even if the velocity is known. Here, we demonstrate a software method to overcome this problem and validate it using high frame rate data with known velocity, which is downsampled to produce aliasing. The method assumes high spatial sampling, constant velocity over short epochs, and an incompressible blood column. Data in successive frames are shifted along the capillary tube axis according to the flow velocity, faithfully rendering cells and plasma. The velocity estimate, required as input to this procedure, can be obtained from either a) the blind optimization of a simple heuristic, or b) a recently proposed velocimetry algorithm, which appears to extend the aliasing limit.