This work constitutes a micromechanical study of ductile fracture.The main methodology consists of acombination of three techniques: Insitu synchrotron radiation computed laminography, digital volume correlation and finite element simulations with advanced meshing and remeshing capabilities. This experimental numerical framework is employed to carry out numerical simulations at the microscale with immersed microstructuresand realistic boundary conditions. The studied material is nodular cast iron, but this methodology has been successfully applied to other materials such as aluminum alloys.Since this is a recent methodology,attention is paid to image immersion as a source of uncertainty. More specifically, the effect of image segmentation on the final mechanical observables, is assessed through comparison of three different segmentation methodologies. The prediction capacity of the numerical simulations is then satisfactorily evaluated through a detailed comparison with experimental measurements. Finally,after evaluating different approaches,the results of the simulations are used to propose and validate a coalescence criterion based on the evolution of the intervoid distance.