3D binder jet technology has drawn significant attention in recent years thanks to its ability to build complex shapes with a wide variety of commercial powders, together with the advantage of not using high energy sources during the printing process. Binder jet makes use of several printheads containing a formulated binder to build the layers of a part within a few seconds. This allows a significant improvement of the manufacturing productivity and quality with respect to laser powder bed fusion (LPBF) or conventional metal injection molding (MIM).Current scientific and industrial investigations in binder jet are focused on the relationship between the printing parameters, the macroscopic properties, and the microstructures of sintered parts. However, there is a knowledge gap related to the relationship between the process parameters and the microstructure and properties of green parts.In this work, a novel green microstructural analysis methodology, based on scanning electron microscopy (SEM) and X-ray computed tomography (XCT), is presented. SEM microstructural observations are supported by machine learning pixel-wise classification algorithms that enables image analysis. This new method facilitates the definition of green parts' key process metrics and the description of consolidation mechanisms (layer consolidation, powder bed interactions) under different printing conditions, before sintering. Thus, the binder and porosity distributions in green microstructures can be correlated to green and sintered macroscopic properties, such as sintered density, with a final modelling and prediction objective for process acceleration. The proposed novel and robust methodology is applied to particular empirical cases.