This study focuses on the influence of brittle secondary phases on the deformation behavior of an Aluminum-Silicon-Iron casting alloy. A digital twin of a coarse grained Al10Si0,7Fe model casting alloy is set up in order to investigate the interaction between the often detrimental ß-Al9Fe2Si2 phase and the silicon phase. The three-dimensional microstructure was determined using computed tomography with an ROI contained in a single grain. Distinction between the Al-matrix and the Si, which possess very similar densities, is possible using a phase contrast scan with the µCT Zeiss Xradia 520 Versa at a resolution of 1.1 µm.  However, the phase boundaries are not very well defined and segmentation after filtering via thresholding yields a high degree of fragmentation of the Si-phase. To further improve the distinction between the Al-matrix and the Si-inclusions the open source Weka machine learning Algorithm was used. Compared to segmentation via thresholding the machine learning algorithm leads to a more realistic connectivity. The reliability of the segmentation procedure is illustrated by comparing the segmentation of an inclusion cluster with SEM images. The 3D model generated by phase sensitive CT analysis combined with machine learning will be used to gain a deeper insight into the role of the inclusion content in the failure behavior of Al-cast alloys.