Microscopy techniques like scanning electron microscopy (SEM) or X-ray computed tomography (CT) can provide detailed image data of electrode particle microstructures. Followed by a quantitative structural characterization such data allows for the investigation of structure-function relationships, i.e., the influence of an electrode particle’s microstructure on its properties like its mechanical or electrochemical behaviour. However, for the structural characterization by means of image data nontrivial processing tasks are often necessary. In this talk several applications of machine learning methods and stochastic geometry are shown for the structural characterization of particles in lithium-ion battery electrodes imaged by SEM, CT and focused ion beam (FIB) - electron backscatter diffraction (EBSD). In the first application, a generative adversarial network (GAN) is deployed to perform super-resolution on SEM-image data of cycled (degraded) cathode particles such that fine features like cracks within particles can be more reliably characterized to investigate the state of degradation. The second application shows how convolutional neural networks can be used to achieve a grain-wise segmentation of FIB-EBSD data of polycrystalline electrode particles. The third application deals with the structural multi-scale modeling of cathode particles to overcome limitations of different imaging techniques. More precisely, a stochastic geometry model is calibrated using both CT data depicting the outer shell of cathode particles and FIB-EBSD data of the grain architecture of a cathode particle. Then, the stochastic model can be used to perform structural scenario analyses, i.e., to generate arbitrarily many digital twins with statistically similar shape and grain architecture as the particles observed in the image data. These digital twins are used as input for numerical charge simulations to investigate their degradation behavior.