In energy research batteries play a pivotal role to account for energy storage. For batteries, the composite materials used in cathode and anode foils are optimized separately and both are decisive for maximum energy density and long-term stability upon cycling. Material optimization for these composites is based on the analysis of scanning electron microscopy (SEM) images, which nowadays can be further enhanced by the incorporation of machine learning (ML). In this study we demonstrate the application of a state-of-the-art machine learning (ML) algorithm to analyze and optimize cathode materials for energy storage batteries. Specifically, a U-Net model is employed here and trained using SEM images of lithium-ion battery cathode foil. The objective is to accurately identify different types of particles, including intact particles, particles with cracks, crushed particles, as well as pores and carbon. The algorithm is trained iteratively, using an initial annotated dataset and subsequent manual annotations assisted by ML. Once the segmentation model is optimized, it becomes possible to extract the individual components from the SEM cathode images and statistical parameters such as particle size distribution, crack density, and porosity can be obtained for further analysis. This information provides insights into the cathode material's structural integrity and quality. The utilization of a segmentation model here offers a more efficient and automated approach compared to labor-intensive and time-consuming manual evaluation methods. The findings presented have the potential to revolutionize the analysis and optimization of cathode materials, benefiting researchers and industry professionals in the development of energy storage systems.