Roughness parameters are traditionally obtained for the surface characterization of microfluidic devices as these are easily computed from height profiles. However, these provide limited topological information on channel geometry needed for fluid flow or fluid retention optimization. In this study, topological measures from Minkowski functionals and cluster analysis were evaluated if these can detect percolating channels or fluid traps in plasma-etched surfaces. Unlike Minkowski functionals, cluster analysis distinguished the onset of microchannel formation from the discontinuous percolation transition. The size of the irregularly shaped microchannel was obtained from the percolation order parameter (P). Thus, P may be used as a topological measure for the automatic detection and geometry evaluation of channels and cavities for various microfluidic applications. Besides percolation, Minkowski functionals and cluster analysis may be used to obtain fractal properties of nano-patterns for nanofluidic device design. (C) 2017 Elsevier B.V. All rights reserved.