Adrenal insufficiency, a life-threatening condition, lacks effective curative treatments. Our recent in vitro protocol successfully differentiated mouse embryonic stem cells into adrenocortical steroidogenic cells without gene overexpression. To enhance efficiency, we want to integrate Deep Learning, which in preliminary data shows promising results in cell type classification. Further optimization will also incorporate molecular data.

Concept of correlating images and molecular data with the goal of predicting the molecular profile of cells from live images. A) Machine learning (ML) algorithm that predicts key genes expressed based on images of cultured cells.  (B) The association of image and gene expression.

We aim to refine differentiation protocols of pluripotent stem cells (PSCs) into adrenocortical steroidogenic cells, following in vivo development. We are fine-tuning 2D culture conditions, investigating specific conditions like extracellular matrix factors, and explore 3D cultures for improved differentiation.. Also we plan to transplant the differentiated cells into mice for in vivo survival and functionality tests.  We are analyzing cell survival, zonation markers, and steroid production, with future functional tests on adrenalectomized and genetically modified mouse models for adrenal insufficiency.