Rigorous quantification of porous microstructures exhibiting a wide variety of pore shapes, sizes, and their spatial distributions presents a significant challenge. In this work, novel data science approaches are used to characterize the complex microstructures in porous membranes, and to extract the salient features at the pore-scale. Towards this goal, a microstructure generator is developed and utilized to create a large ensemble of porous structures covering a substantial range in measures of features such as the stretched pore shapes (geometrical anisotropy), porosity, specific surface, and pore sizes. Additionally, the morphology of real porous membranes are obtained experimentally by high resolution X-ray tomography. The statistical representations for the simulated and real membrane microstructures are calculated and compared rigorously using novel data science approaches that are based on principal component analyses of the 2-point spatial correlations. This approach allows an objective measure of the difference between any two selected microstructures. The versatility and benefits of this approach for the quantification of microstructures in porous membranes are demonstrated in this paper.