HMN 2025: How Assessing blood stem cell quality by analyzing cell behavior in real time

Hematopoietic stem cells (HSCs) are the fundamental building blocks of our circulatory system, giving rise to all blood cell types, including white blood cells, red blood cells, and platelets. HSCs play a key role in our understanding of complex biological processes and are involved in life-saving treatments such as bone marrow transplants and emerging gene therapies. As a cornerstone of regenerative medicine, HSCs hold immense promise for treating blood disorders, cancers, and immune system diseases.

However, the use of HSCs in clinical applications faces complex challenges. Accurately predicting the long-term quality of individual HSCs is crucial for cell-based therapies, yet this remains difficult. A core limitation lies in the use of “snapshot” analytical methods, which capture a cell’s state at only a single instant in time. These static assessments cannot account for the dynamic changes that define HSC behavior and differentiation, ultimately overlooking important characteristics that may influence their therapeutic potential.

To address these challenges, a research team led by Assistant Professor Takao Yogo and Professor Satoshi Yamazaki from The Institute of Medical Science at The University of Tokyo, Japan, developed an innovative approach to predict HSC functionality. The researchers combined advanced imaging technology with machine learning to create a system that is capable of predicting HSC quality based on real-time cellular behavior. This study was published in the journal Nature Communications on July 14, 2025.

The team’s innovative approach integrates single-cell expansion cultures with quantitative phase imaging (QPI), a cutting-edge technique that enables continuous observation of living cells without damaging them or using fluorescent markers. By capturing detailed videos of individual HSCs lasting up to 96 hours, the researchers were able to extract a wealth of information about their cellular kinetics—how they move, grow, and divide.

A key finding of their study was the discovery of previously hidden diversity within HSC populations. Even among cells that appeared identical under traditional snapshot analyses, the QPI time-lapse videos revealed variations in proliferation rates, shapes, motility, and division patterns.

This so-called “infinite diversity” implies that HSCs are far more complex and dynamic than previously understood. Moreover, the team also demonstrated that kinetic features observed through QPI could successfully predict the expression levels of Hlf, a gene that serves as a reliable indicator of the “stemness” (and thus quality) of HSCs.

The researchers also developed and trained a deep neural network with their HSC time-dependent data, allowing it to predict Hlf expression from captured cellular kinetics. Notably, the predictive power of this system dramatically improved when more temporal information was included, highlighting the advantages of the proposed approach over snapshot analyses.

Overall, the innovative strategy proposed in the study could help push the boundaries of our knowledge about HSCs and their unique characteristics. “This breakthrough allows for a scientific analysis of previously inaccessible cell populations and is expected to catalyze advancements in basic biology and technological innovation in stem cell science,” remarks Dr. Yogo.

Additionally, the new approach represents a significant step forward in medical applications. In regenerative medicine and gene therapy, the quality of transplanted cells is paramount, and there have been reports of adverse effects potentially caused by low-quality HSCs.

“Our system enables a new level of cell quality control,” notes Dr. Yogo. “Ultimately, this research establishes a new analytical paradigm in the fields of gene therapy and regenerative medicine, serving as a foundational technology that directly contributes to the safety and effectiveness of future cellular therapies.”