Cells can now be filmed as they are, in 3D, with no dye added

A University of Tokyo team has built a microscope that records chemically specific 3D images of living cells 19.2 times a second, nearly 400 times faster than earlier versions of the technique. ^[1][2]

The study, led by first author Masato Fukushima with Takuro Ideguchi as senior author, appeared in Nature Communications on October 30, 2026. ^[1] It is a peer-reviewed laboratory demonstration. It shows that the method works and how fast it runs. It does not yet show what the method will find.

From slide show to film

A cell is a crowded, restless place, and most of the ways we look at it are slow. Take a 3D picture of the inside of a cell and you must collect many views and stitch them together. By the time the last view is in, the first is out of date.

The method the Tokyo team improved is called mid-infrared photothermal tomography. Earlier wide-field versions managed fewer than 0.05 volumes per second, according to the study. ^[1] That is no faster than one 3D image every 20 seconds. The new instrument produces one every 52 milliseconds. ^[1]

A lot can happen inside a cell in 20 seconds. Things that move, fuse and drift apart in that time showed up in older instruments as a blur or not at all. At 19.2 volumes per second the same motion becomes a sequence of distinct frames.

How it sees

Molecules absorb mid-infrared light at wavelengths that act as chemical signatures, so different molecules can be told apart by what they soak up. When a molecule absorbs the light it warms slightly, and the warmth changes how much the material bends light, a property called the refractive index. Heat shimmer over a road works on the same principle, since warmed air bends the light passing through it. ^[1]

The microscope does not look at the infrared light itself. It probes the cell with visible light and measures that tiny bending. It takes readings with the infrared on and off, and from several angles, then reconstructs a 3D chemical image from them. ^[1]

The speed comes from fixing two bottlenecks the study names. Earlier systems got too little signal from each viewing angle, or took too long to switch between angles. ^[1]

The Tokyo instrument uses a pulsed infrared source that delivers enough signal in a single camera exposure, with no need to average many exposures. ^[1] A high-speed spatial light modulator, a device that can reshape and steer light, flips the visible probe between 11 angles. ^[1]

The authors describe a signal-to-noise ratio above 70 under video-rate conditions. ^[1][2] In plain terms, the signal stands well clear of the static.

Fat droplets on the move

For a first demonstration the team chose lipid droplets, the small fat-storing bodies found in cells. They tracked the droplets through living cells and measured their paths in three dimensions. ^[1]

The droplets did not all behave alike. The study reports varied transport behavior that two-dimensional measurements obscured. ^[1] Flatten a 3D journey onto a plane and different paths can look the same. That is the practical case for doing this in volume and at speed.

A second mode trades volume for spectrum. By sweeping the infrared wavelength rapidly, the team collected 20 spectral points across a 300 inverse-centimeter window in about one second. ^[1]

Inverse centimeter is the unit infrared spectroscopists use to mark a position on the spectrum. It counts how many wave crests fit into one centimeter, so a bigger number means a shorter wavelength. A 300 inverse-centimeter window is a stretch of the infrared 300 units wide on that scale, and across it the molecules in a cell each absorb at their own spots.

The signatures were enough to tell apart structures including nucleoli and lipid droplets. ^[1] A nucleolus is the dense region inside a cell's nucleus where the cell's protein-making machinery is assembled.

Why no dye matters

Most live-cell movies today rely on fluorescent labels, molecules added to a cell so that chosen parts glow. The authors note that labels can perturb the biological state being studied or complicate sample preparation. ^[1] Vibrational imaging, the family this method belongs to, reads chemical information from molecules already in the cell. ^[1]

So the pitch is a film of a cell as it is, rather than a cell dressed for the camera. The trade is that the signal is weaker and harder to get, which is why the speed record is a real result and not a cosmetic one.

The researchers name possible uses ahead: imaging how a drug accumulates in a cell, how lipid droplets form and fuse, and how aggregates build up inside cells. ^[1] These are prospects, not findings. The study does not test cancer-cell migration or the firing of neurons, two applications that tend to be mentioned whenever fast cell imaging comes up.

What is still open

The published description gives no count of cells or biological replicates, so how reliably the droplet behavior repeats across samples is not yet clear. Also untested in public are how the image quality holds up in denser samples or longer sessions, whether the infrared light and local heating harm cells during extended imaging, and whether the instrument can leave the lab that built it. Those are the questions other groups will ask first.

The work has been visible for some time. The authors posted a preprint, an early version released before peer review, on arXiv on March 4, 2026. ^[2] The journal paper is the version that has passed review.

What to watch next is whether anyone points this instrument at something that matters to a patient, such as a drug entering a cell, and whether the cells survive the experience.