Pluripotent stem cells can be used for both regenerative medicine and in vitro modeling for diseases and drug screening. For regenerative medicine, pluripotent stem cells need to differentiate into a specific cell type that is required for transplantation therapy. Cell products for transplantation therapy have lot-to-lot variations in quality, which affect their efficacy and side effects as well as the cost of the products. Quality evaluation of cell products requires non-invasive methods although well-used methods, such as expression analysis of genes and proteins, deconstruct cell products. To evaluate the quality of cell products in a non-deconstructive way, we developed a machine learning model that predicts the function of cell products using label-free images of cell morphology. Next, for in vitro modeling, we sought to recapitulate the interaction of brain regions by fusing brain-region-specific organoids. Imaging and viral vectors allowed for the evaluation of in vitro interaction of brain regions. In this talk, we will introduce our recent work with multidisciplinary approaches.