A research team from the Institute of Medical Science at the University of Tokyo in Japan has published innovative results in Nature Communications, developing a quality prediction system for hematopoietic stem cells based on analysis of their dynamic behavior. This technology combines quantitative phase imaging with deep learning and is expected to improve the safety and efficacy of cell therapy.

Traditional assessment of hematopoietic stem cells relies on static "snapshot" methods, which fail to fully capture cell characteristics. Assistant Professor Takao Yogo from the research team stated: "Our system elevates cell quality control to an entirely new level." The new technology continuously observes cells over 96 hours, capturing dynamic features such as stem cell movement and proliferation, revealing cell diversity that traditional methods cannot identify.

The study employed quantitative phase imaging technology to record changes in hematopoietic stem cell behavior without damaging the cells. Combined with deep learning algorithms, the system can accurately predict the expression level of the Hlf gene—a key indicator for evaluating stem cell quality. "This breakthrough enables scientific analysis of previously inaccessible cell populations," Dr. Yogo added.

This technology represents a significant advancement in the field of cell therapy, particularly in applications such as bone marrow transplantation and gene therapy. The researchers noted that precise quality prediction can avoid adverse reactions caused by low-quality cells and improve treatment success rates. The team plans to further optimize the system to promote its practical application in clinical settings.