Reading blood with light: label‑free imaging teases apart immune cell behavior
What can we learn about the immune system from a plain vial of blood—without adding a single label or dye—and how soon after a stimulus do those changes appear? The study asked whether a purely optical read on cell metabolism can sort different immune cells and tell who is “on” or “off” inside a mixed blood sample.
The work comes from researchers in the Skala Lab, with senior author Melissa Skala of the Morgridge Institute for Research and the University of Wisconsin–Madison, and appears in Biophotonics Discovery as Autofluorescence lifetime imaging resolves cell heterogeneity within peripheral blood mononuclear cells.
Why PBMCs, and why metabolism?
Peripheral blood mononuclear cells (PBMCs) are the workhorse of human immunology: a routine draw yields a diverse mix used to study infections, autoimmune disorders, cancer, and how treatments are working. They’re also clinically straightforward to isolate. But the usual way to tell who’s who—fluorescent labels stuck to surface markers—doesn’t say much about how the cells are actually functioning and can alter the cells during prep.
The study’s bet was that metabolism carries that missing information. If you can read how cells are making and using energy without touching them, you might see which ones are activated and which subset you’re looking at, inside the full PBMC mix.
Method: label‑free optical metabolic imaging inside mixed samples
The researchers used optical metabolic imaging (OMI), a technique developed and refined by the Skala Lab. Crucially, it is nondestructive and measures metabolism in individual immune cells while they remain in a heterogeneous PBMC sample—no external dyes or cell sorting first.
They applied OMI to PBMCs isolated from the blood of three healthy donors, profiling cells in resting conditions and after activation. The analysis asked a simple classification question: do metabolic measurements alone distinguish activation state and identify key cell populations inside that unseparated mix?
What they found, in numbers
Metabolism turned out to be a strong signal. Using only the metabolic readout, the team distinguished activated from resting PBMCs with nearly 94% accuracy just two hours after stimulation. Monocytes were identified with 96% accuracy in resting samples and 88% in activated samples. Natural killer (NK) cells were identified at approximately 74% accuracy in both states.
Those results line up with a broader observation in the data: monocytes and NK cells show particularly distinct metabolic signatures inside mixed PBMCs. That tracks with their roles as rapid responders, where big shifts in energy use mark cells that are gearing up for action.
| Question | Result / Detail |
|---|---|
| Sample source | PBMCs from 3 healthy donors |
| Approach | Optical metabolic imaging (label-free, nondestructive) |
| Activation readout | Activated vs. resting PBMCs distinguished with ~94% accuracy 2 hours after stimulation |
| Monocyte identification | 96% accuracy (resting); 88% accuracy (activated) |
| NK cell identification | ~74% accuracy (resting and activated) |
| Key pattern | Monocytes and NK cells exhibit distinct metabolic signatures |
What this enables right now
Because the measurements leave cells intact, the same PBMCs remain available for further analyses or potential therapeutic use. That matters for research programs that already lean on PBMCs to track disease course or treatment response, where a quick, nondestructive window into activation could complement existing counts and labels.
It also fits with a wider push to pair advanced biophotonic imaging with machine learning to make metabolic information more accessible in studies of cancer, immune disorders, and cell therapies.
The deeper cut
How to read “accuracy” in a mixed-cell problem
When a study reports ~94% accuracy for separating activated from resting PBMCs using metabolism alone, that metric reflects a binary classification across a heterogeneous pool. In such mixes, class imbalance and overlap matter: a small activated fraction that is metabolically extreme can drive high accuracy even if borderline cases blur. Likewise, the monocyte (96% resting; 88% activated) and NK (~74%) figures are effectively separability scores for those phenotypes under the chosen activation condition and time point. They say that the metabolic feature space forms more compact, better-separated clusters for monocytes than for NK cells—consistent with distinct metabolic signatures in the former.
Practically, the useful comparison is not to perfection but to label-based baselines. Established marker panels can discriminate many subtypes with higher fidelity. The value here is orthogonality: a label-free axis of information that reflects function. In workflows, that can act as a prescreen (e.g., flag “metabolically fit” fractions) or as a sanity check against marker-defined gates. The main caveats are stability over donors and conditions, and drift over time post-stimulation—the study’s two-hour snapshot is favorable for activation contrast but may not generalize to other windows without retraining or recalibration.
One section on what we don’t know (yet)
There are clear limits. The team worked with PBMCs from three healthy donors, and they emphasize the technology is primarily a research tool. It does not yet match the accuracy of established labeling methods for identifying every immune-cell subtype.
The results are anchored to specific conditions, including the two-hour post‑stimulation window. How performance changes with other stimuli, time points, diseases, or broader subtype panels remains to be mapped before clinical adoption.
Why the approach is attractive anyway
PBMCs are easy to obtain and already central to studies of infection, autoimmunity, cancer, and treatment response, so adding a nondestructive functional read could sharpen those tools without changing the clinical workflow. The distinct metabolic fingerprints seen for monocytes and NK cells point to immediate use cases where innate responses are the question at hand.
Looking ahead, combining biophotonic imaging with machine learning is the path the researchers highlight for turning these metabolic snapshots into accessible, scalable assays for cancer research, immune disorders, and cell therapies.
The paper: Autofluorescence lifetime imaging resolves cell heterogeneity within peripheral blood mononuclear cells (Biophotonics Discovery, 2026)
Sources: Advanced optical imaging reveals hidden activity in blood immune cells (phys.org)
Images: Cover: Authors of the study: Jiabao Xu, Tiffany Lodge, Caroline Kingdon, James W. L. St / Wikimedia Commons (CC BY 4.0)
How this article was made: Everyman Science uses AI tools to structure, format and optimise its articles, and occasionally to produce illustrations where no free photograph exists. The reporting these articles are based on is human-produced and cited above. Spotted an error? Write to [email protected] and we will correct it. — The editors How we work.
Science desk team of Everyman Science, curating and reporting on the day’s most significant developments in research, space exploration, and technology.
