Image analysis measurements of roughness by texture and fractal analysis correlate with contact profilometry

Surface properties of a biomaterial are important factors that govern in part its biocompatibility. Among them, surface roughness is now recognized as a very important factor for cell interactions. Surface roughness (Ra) is routinely measured by contact profilometry but other methods are presently usable. We compared two methods (contact profilometry and image analysis of scanning electron microscopic images SEM) on a series of 12 titanium test pieces. The texture analysis of SEM pictures was done by the heterogeneity and run-length methods. Fractal geometry was also used with the "skyscraper" and "blanket" methods providing respectively the D-SKY and D-BLANK fractal dimensions. The fractal dimension of the profilometric curve was also computed (D-MINK). Computer-simulated textures were used to evaluate the pertinence of the algorithms. A significant correlation was found between Ra and all the texture descriptors except heterogeneity. The correlation coefficient was dependent on the microscopic magnification. The fractal dimension of the curve was correlated with D-SKY and D-BLANK. Run-length, D-SKY and D-BLANK were highly correlated, independent of the magnification used, a finding related to the self-similarity of the images. Image texture analysis can be a useful alternative to profilometry with brittle or soft materials or with objects having a complex shape. (C) 2002 Elsevier Science Ltd. All rights reserved.