The aim of this paper is to propose a new approach permitting the investigation of the 3D pore variation of kaolin K13 in relation with mechanical loading. A mechanical loading consisting of one-dimensional compression and oedometric loading is applied to a sample of centimeter scale. A postmortem protocol using scanning electron microscopy (SEM) coupled with Focused Ion Beam (FIB) was carried out to observe and then obtain a large number of slices of an extracted sub-volume from the small sample. Image analysis using a new processing method enabled the reconstruction of the observed and extracted sub-volume in the form of a rebuilt sub-volume. The segmentation of pores and particles was carried out using an approach based on machine learning and the pore space properties can be quantified. Pore morphology was identified based on two parameters namely the flatness and the elongation. Spatial orientation can also be locally determined. The proposed treatment enabled the distribution of pore size (termed Im_PSD) to be deduced and these were then compared with the results of Mercury intrusion porosimetry technique.