The cerebral organoid is an emerging model system with high potential for both fundamental and applied neuroscience research. However, batch-to-batch variability and the inability to characterize these specimens at the cellular level with high throughput, hampers their integration in an industrial setting. With this project, we intend to develop a pipeline that enables unbiased cellular phenotyping of intact cerebral organoids by using a combination of multiplex fluorescent labelling, light-sheet microscopy, and deep learning. Our approach builds on the concept that cells can be accurately identified by means of sheer morphological information. First, we will perfection cell profiling in co-cultures of different brain cell lines by training machine learning-based classifiers. Then, we will translate the concept to 3D, using spheroids from the same cells. To this end, we will render the staining compatible with chemical clearing and conceive a sample mounting procedure for serial, isotropic image acquisition. Finally, we will deploy the optimized method to recognize cell type and state in iPSC-derived cerebral organoids that have been challenged with selected compounds or seeded with glioblastoma cells. The approach will add to a more standardized quality-control of organoids and will facilitate the adoption of this model in drug-screening pipelines. Ultimately, this will boost the translatability of preclinical research and lower attrition rates in late-stage clinical trials.