Automated fiber diameter and porosity measurement of fibrin clots in SEM images

Abstract

SEM images of fibrin meshes can provide profound insights into the microstructure of blood clots. This is significant because characteristics of clot microstructure, such as fiber diameter and clot porosity, may be altered in disease (e.g., cardiovascular disease) and by other factors (e.g., medications). In the past, SEM images of fibrin clots were analyzed manually - a burdensome and bias-prone process, in which a user inspects images and makes measurements manually. Automated measurements have been lacking due to the absence of finely tunable and accurate software to extract data from the SEM images. Recently, a plug-in for ImageJ (open source, NIH program), called DiamterJ, was developed for analyzing SEM images of porous networks. We evaluated the suitability of DiameterJ for diameter and porosity measurements on up to 150 SEM images of fibrin meshes by comparing the outcomes with manual measurements and correlating them to biophysical properties of the clot. Three out of 24 segmentation algorithms (S4, S8, M4) were suitable for diameter determination, yielding diameter values that strongly correlated with, and were within 8% of, manually determined values. However, all three algorithms left gaps in the images. A different algorithm (T5), which closely overlaid with all fibers in the images, maybe suitable to compute porosity and pore area from SEM images; though this task remains challenging due to uncertainties in clot depth. Manually determined diameter, and diameter as determined by the S4, S8 and M4 algorithms, correlated strongly to maximum absorbance, moderately to fibrinogen concentration, but not to permeability, storage modulus G', loss modulus G'' and G'/G''. Three algorithms in DiamterJ provide reliable, automated fiber diameter measurements in fibrin clots, other algorithms overestimate the diameter. Automated porosity and mean pore size determination remain challenging