Visually Identifying Senescent Cells with an Algorithm

Details of a cell's size and shape might offer more information than other biomarkers.


Cells Under MicroscopeCells Under Microscope

A team of researchers publishing in Aging has developed a method of identifying senescent cells through their physical morphology, potentially making future senescence research much easier.

Senescent cells look different

Before modern biomarkers of senescence were eludicated, cell size was considered to be one of its defining features, and the researchers cite a 1969 paper to that effect [1]. This excessive growth has been shown to have a causal relationship with senescence, with larger cells having more diluted cytoplasm that contributes to the condition [2]. Other prior work has found links between senescence and nuclear size and shape [3], and a common staining method shows less visual intensity in senescent cells, as they tend to be flatter under a microscope [4].

These visual differences have largely fallen out of favor among the research community as identifiers, being replaced by well-known chemical biomarkers. However, modern image processing algorithms, specifically the high-content analysis (HCA) used in this study, may make it easier to look at senescent cells instead of biochemically analyzing them.

An analysis with a lot of variables

Throughout this study, the researchers primarily examined four different populations of senescent cells: a line of human fetal lung fibroblasts with oncogene-induced senescence (OIS), the same line of fibroblasts with SASP-induced senescence, human mammary fibroblasts (HMFs) with replicative senescence, and human dermal fibroblasts (HDFs) with replicative senescence.


Eterna is a clothing company with a focus on longevity.

The researchers also used 62 separate biomarkers to determine the differences between proliferating and senescent cells, based both on Z-score analysis, which uses defined controls as reference points, and standard normalization, which uses averages from the given sample. In an effort to simplify future analyses, the researchers also used an algorithmic technique called exploratory factor analysis to generalize these 62 factors into eight “latent factors”, such as nucleus size and cell shape.

Heterogeny and morphology

There were substantial differences between the types of senescent cells. Cells that had reached their replicative limit and became senescent due to telomere attrition were less visually distinguishable from their proliferating counterparts, and the researchers attribute this to the gradual nature of replicative senescence.

Cells that became senescent through oncogenes and cells that became senescent due to the SASP had more substantial differences from proliferating cells. However, these two groups also differed somewhat from each other: the OIS cells had more differences in cell shape, while the SASP-induced group had more differences in visual intensity.

There were also differences from the cells within each model. Cells in OIS group were emitting the SASP to each other, and the researchers hold that some cells may have been more affected by this process than by the original oncogenes, which would logically lead to variances in the cells’ morphology.

Visualizing the data through heat maps corroborated these findings. The OIS and SASP-influenced groups were in distinct clusters from proliferating cells, while the replication-induced groups’ results had fuzzier borders.


An advertisement banner for PartiQular supplements.

An in vivo analysis

The researchers also confirmed their findings through an in vivo analysis. They grew tumors of human cells in mice, induced senescence with a drug, and tested the cells for the senescence biomarker p21 and examined their cellular morphology. The researchers’ algorithms reported 27 substantial differences in morphology between cells that were positive and negative for this biomarker.


Image analysis is continuously being refined as a diagnostic tool in medicine, and visual identification simply requires placing the cells under a suitable microscope rather than biochemistry of any kind. Using physical differences might let researchers better distinguish between the types of senescent cells, possibly leading to a greater understanding of the nature of these cells. If such an algorithm can be refined and proven reliable, it might become a new, superior standard for the identification of senescent cells.

To do this, we need your support. Your charitable contribution tranforms into rejuvenation research, news, shows, and more. Will you help?


[1] Cristofalo, V. J., & Kritchevsky, D. (1969). Cell size and nucleic acid content in the diploid human cell line WI-38 during aging. Pharmacology, 19(6), 313-320.

[2] Neurohr, G. E., Terry, R. L., Lengefeld, J., Bonney, M., Brittingham, G. P., Moretto, F., … & Amon, A. (2019). Excessive cell growth causes cytoplasm dilution and contributes to senescence. cell, 176(5), 1083-1097.

[3] Sadaie, M., Dillon, C., Narita, M., Young, A. R., Cairney, C. J., Godwin, L. S., … & Narita, M. (2015). Cell-based screen for altered nuclear phenotypes reveals senescence progression in polyploid cells after Aurora kinase B inhibition. Molecular biology of the cell, 26(17), 2971-2985.


[4] Zhao, H., Halicka, H. D., Traganos, F., Jorgensen, E., & Darzynkiewicz, Z. (2010). New biomarkers probing depth of cell senescence assessed by laser scanning cytometry. Cytometry Part A, 77(11), 999-1007.

About the author
Josh Conway

Josh Conway

Josh is a professional editor and is responsible for editing our articles before they become available to the public as well as moderating our Discord server. He is also a programmer, long-time supporter of anti-aging medicine, and avid player of the strange game called “real life.” Living in the center of the northern prairie, Josh enjoys long bike rides before the blizzards hit.