A combination of high-resolution imaging and machine learning, also known as artificial intelligence (AI), can track cells damaged from injury, aging, or disease, and that no longer grow and reproduce normally, a new study shows.
These senescent cells are known to play a key role in wound repair and aging-related diseases, such as cancer and heart disease, so tracking their progress, researchers say, could lead to a better understanding of how tissues gradually lose their ability to regenerate over time or how they fuel disease. The tool could also provide insight into therapies for reversing the damage.
Led by researchers at NYU Langone Health’s Department of Orthopedic Surgery, the study included training a computer system to help analyze animal cells damaged by increasing concentrations of chemicals over time to replicate human aging. Cells continuously confronted with environmental or biological stress are known to senesce, meaning they stop reproducing and start to release telltale molecules indicating that they have suffered injury.
Published in the journal Nature Communications online July 7, the researchers’ AI analysis revealed several measurable features connected to the cell’s control center (its nucleus) that when taken together closely tracked with the degree of senescence in the tissue or group of cells. This included signs that the nucleus had expanded, had denser centers or foci, and had become less circular and more irregular in shape. Its genetic material also stained lighter than normal with standard chemical dyes.
Further testing confirmed that cells with these characteristics were indeed senescent, showing signs that they had stopped reproducing, had damaged DNA, and had densely packed enzyme-storing lysosomes. The cells also demonstrated a response to existing senolytic drugs.
From their analysis, researchers created what they term a nuclear morphometric pipeline (NMP) that uses the nucleus’s changed physical characteristics to produce a single senescent score to describe a range of cells. For example, groups of fully senescent cells could be compared to a cluster of healthy cells on a scale from minus 20 to plus 20.
To validate the NMP score, the researchers then showed that it could accurately distinguish between healthy and diseased mouse cells from young to older mice, age 3 months to more than 2 years. Older cell clusters had significantly lower NMP scores than younger cell clusters.
The researchers also tested the NMP tool on five kinds of cells in mice of different ages with injured muscle tissue as it underwent repair. The NMP was found to track closely with changing levels of senescent and nonsenescent mesenchymal stem cells, muscle stem cells, endothelial cells, and immune cells in young, adult, and geriatric mice. For example, use of the NMP was able to confirm that senescent muscle stem cells were absent in control mice that were not injured, but present in large numbers in injured mice immediately after muscle injury (when they help initiate repair), with gradual loss as the tissue regenerated.
Final testing showed that the NMP could successfully distinguish between healthy and senescent cartilage cells, which were 10 times more prevalent in geriatric mice with osteoarthritis than in younger, healthy mice. Osteoarthritis is known to progressively worsen with age.
Our study demonstrates that specific nuclear morphometrics can serve as a reliable tool for identifying and tracking senescent cells, which we believe is key to future research and understanding of tissue regeneration, aging, and progressive disease.”
Michael N. Wosczyna, PhD, study senior investigator
Dr. Wosczyna is assistant professor in the Department of Orthopedic Surgery at NYU Grossman School of Medicine.
Dr. Wosczyna says his team’s study confirms the NMP’s broad application for study of senescent cells across all ages and differing tissue types, and in a variety of diseases.
He says the team plans further experiments to examine use of the NMP in human tissues, as well as combining the NMP with other biomarker tools for examining senescence and its various roles in wound repair, aging, and disease.
The researchers say their ultimate goal for the NMP, for which NYU has filed a patent application, is to use it to develop treatments that prevent or reverse negative effects of senescence on human health.
“Our testing platform offers a rigorous method to more easily than before study senescent cells and to test the efficacy of therapeutics, such as senolytics, in targeting these cells in different tissues and pathologies,” said Dr. Wosczyna, who plans to make the NMP freely available to other researchers.
“Existing methods to identify senescent cells are difficult to use, making them less reliable than the nuclear morphometric pipeline, or NMP, which relies on a more commonly used stain for the nucleus,” said study co-lead investigator Sahil Mapkar, BS. Mapkar is a doctoral candidate at the NYU Tandon School of Engineering.
Funding for the study was provided by National Institutes of Health grant R01AG053438 and the Department of Orthopedic Surgery at NYU Langone.
Besides Dr. Wosczyna and Mapkar, NYU Langone researchers involved in this study are co-lead investigators Sarah Bliss and Edgar Perez Carbajal and study co-investigators Sean Murray, Zhiru Li, Anna Wilson, Vikrant Piprode, Youjin Lee, Thorsten Kirsch, Katerina Petroff, and Fengyuan Liu.
Source:
Journal reference:
Mapkar, S. A., et al. (2025). Nuclear morphometrics coupled with machine learning identifies dynamic states of senescence across age. Nature Communications. doi.org/10.1038/s41467-025-60975-z.
