Image optimization and analysis of synchrotron X-ray computed microtomography (CμT) data

Synchrotron X-ray computed microtomography (CmuT) is a non-destructive technique for imaging porous and compositionally heterogeneous samples in three dimensions at the microscale. In this study, we report a package of FORTRAN algorithms for digital image optimization and three-dimensional analysis of porosity, pore connectivity, and pore structure within a CmuT volume. The algorithm Tomo_optimize optimizes digital data by utilizing a series of matrix filters and contrast transforms. Tomo_classify labels individual voxels within the data set as solid, internal pore space. or external void space, thus defining virtual volume boundaries. Tomo_analysis calculates total porosity, porosity from interior pores (completely surrounded by solid), and porosity from connected pores (open to external void space), and provides an output of each pore and its pore size (number of voxels per pore). The algorithms were tested on two natural samples from hydrothermal vent chimneys. Physical volume was 116 and 72 mm(3) for each sample and CmuT spatial resolution was estimated to be 57 mum. Porosity determined by the CmuT algorithms was 14.1% and 15.4%, respectively. The majority of porosity (> 98 %) was due to connected pores rather than isolated pores, and most of the pore volume contributing to total porosity of both samples (> 90%) was from one large interconnected pore. While total porosity was similar for both samples, three-dimensional visual reconstructions showed a more channelized pore structure in one sample. Sensitivity analyses were performed to test the effect of different cut-off values for air, internal pore space, and solid entered by the user before and after image processing on porosity calculations. These algorithms provide an integrated image processing and analysis package for synchrotron CmuT data that should be useful for the analysis of microporous structures as this technique gains popularity. (C) 2003 Elsevier Science Ltd. All rights reserved.