MIT develops AI-powered Raman microscopy to detect aging cells

MIT develops AI-powered Raman microscopy to detect aging cells

Researchers at MIT have created a noninvasive method to identify senescent cells—the so-called "zombie cells" that stop dividing but refuse to die as we age. The breakthrough combines Raman microscopy with single-cell gene expression analysis to generate unique molecular "barcodes" that can quickly flag these cells in tissue samples.

The work, first reported by MIT News, could eventually enable doctors to diagnose and monitor age-related conditions by detecting cellular senescence inside the body using endoscopic tools.

Why it matters

Senescent cells accumulate as the immune system's cleanup mechanisms weaken with age, contributing to conditions ranging from sagging skin and muscle weakness to osteoarthritis, type 2 diabetes, and cancer. Current detection methods require destroying the cells being studied. A noninvasive diagnostic would let clinicians track senescence in living tissue over time and evaluate whether emerging anti-aging therapies actually clear these cells.

How the technique works

Raman microscopy reveals a cell's chemical composition by analyzing how it scatters near-infrared or visible light—without harming the sample. The MIT team paired this optical technique with spatial RNA sequencing, which maps where genes are active within tissue, to build a comprehensive profile of senescent cells.

Working with skin and lung tissue from young (2-month-old) and old (26-month-old) mice, the researchers identified dramatic biochemical shifts in aging cells. The most striking change was a sharp increase in lipid synthesis and accumulation across both tissue types. Senescent skin cells also showed altered pathways for muscle contraction and collagen remodeling, while aged lung tissue exhibited heightened immune activation and inflammation.

By correlating specific Raman spectral peaks—signatures of particular chemical bonds—with gene expression patterns, the team identified combinations that reliably indicate senescence. These peaks correspond to lipids, proteins, and other molecules that change as cells enter their zombie state.

From lab to clinic

The current system requires roughly 30 hours to analyze a one-square-millimeter tissue sample. The researchers are now engineering a faster version that can rapidly scan larger areas for the Raman barcodes they've identified.

Jeon Woong Kang, an MIT research scientist and senior author on the study, envisions future endoscopes equipped with this technology to detect cellular senescence inside patients' bodies. The team is adapting the mouse-based method for use with human tissue.

The research is part of the National Institutes of Health's Cellular Senescence Network, an initiative aimed at understanding how senescent cells contribute to both normal physiology—including embryonic development and tissue repair—and disease processes.

Peter So, director of MIT's Laser Biomedical Research Center and professor of biological engineering and mechanical engineering, and Jian Shu of Massachusetts General Hospital, Harvard Medical School, the Broad Institute, and the Ragon Institute, served as senior authors. The findings appear in Nature Aging and were funded by the NIH and Massachusetts General Hospital.