The two-dimensional data set is azimuthally integrated to produce one-dimensional patterns. The data representation S(y, ω) required for the reconstruction is the global sinogram. All sinograms have been represented before direct beam intensity normalization. Different regions of interest (ROIs), corresponding to phases differentiated through their d spacing, are selected: in this example (a) capillary, (b) crystalline diamond, (c) amorphous carbon and (d) graphite-like carbon. The reconstruction provides the intensity distribution as a two-dimensional image (x, y) for each ROI [a detailed analysis for this sample is given by Álvarez-Murga et al. (2012)].