Discussion: “Comparison of Statistical Methods for Assessing Spatial Correlations Between Maps of Different Arterial Properties” (Rowland, E. M., Mohamied, Y., Chooi, K. Y., Bailey, E. L., and Weinberg, P. D., 2015, ASME J. Biomech. Eng., 137(10), p. 101003): An Alternative Approach Using Segmentation Based on Local Hemodynamics

The biological response of living arteries to mechanical forces is an important component of the atherosclerotic process and is responsible, at least in part, for the well-recognized spatial variation in atherosusceptibility in man. Experiments to elucidate this response often generate maps of force and response variables over the arterial surface, from which the force–response relationship is sought. Rowland et al. discussed several statistical approaches to the spatial autocorrelation that confounds the analysis of such maps and applied them to maps of hemodynamic stress and vascular response obtained by averaging these variables in multiple animals. Here, we point out an alternative approach, in which discrete surface regions are defined by the hemodynamic stress levels they experience, and the stress and response in each animal are treated separately. This approach, applied properly, is insensitive to autocorrelation and less sensitive to the effect of confounding hemodynamic variables. The analysis suggests an inverse relation between permeability and shear that differs from that in Rowland et al. Possible sources of this difference are suggested.