Micro-particle image velocimetry measurement of blood flow: validation and analysis of data pre-processing and processing methods

The intent of this paper is to investigate the application of a pre-processing method previously validated on glycerol to blood flows in microchannels and to compare the accuracy of results obtained when applied to a non-homogeneous fluid such as blood with results from previously applied processing methods for blood data. Comparisons of common processing methods are desired for a clear measure of accuracy in order to make recommendations for various flows. It is hypothesized that increasing the correlation window overlap improves the profile prediction. The amount of correlation window overlap and window shape in the processing of data have a significant effect on the results. Image pre-processing is explored to improve the correlation using the 'image overlapping' which is extended to the case of blood and the blood-specific pre-processing 'base-clipping' or 'thresholding' technique currently applied to blood. Both pre-processing methods are tested with multiple processing methods for two channel geometries: a straight rectangular channel and a Y-channel resulting in a controlled shear flow. The resulting profiles and calculations demonstrate that 'image-overlapping' is found to achieve a profile closer to the predicted theoretical profile than current blood pre-processing methods when both are applied to the same set of data and both are superior to conventional cross-correlation on its own. In all cases, pre-processing decreases the smoothness of the predicted profile. The use of 'image-overlapping' is shown to have greater accuracy when calculating the shear rate at the wall of the channel as well. (paper)