Representative fluorescence microscopy images of cultured cells. (a) Nuclei of HeLa cells stained with DAPI, imaged on a confocal high-content screening system at 40×. (b) HeLa cells with a fluorescent protein targeted to the Golgi apparatus (orange), microtubules (green), and DNA (cyan), imaged by multiphoton microscopy. (c) Mitochondria (red) in a cultured cell. Each color is one label, recorded as a separate channel and pseudocolored for display. Image credits: (a) DrBib, CC BY 4.0; (b) National Institutes of Health; (c) NICHD/NIH; (b) and (c) public domain; all via Wikimedia Commons.
In 2.8, we saw how phase-contrast and DIC microscopy make unstained cells visible. These images reveal the shape and position of cells, but not their molecular composition or chemistry. Consider two cells that look almost identical in phase contrast, one dividing and one not. To distinguish them, we need to identify the proteins involved in cell division. Fluorescence microscopy makes this possible.
Instead of relying on how the whole specimen alters light, fluorescence microscopy relies on light-emitting molecules attached to specific targets, such as proteins, DNA sequences, or organelles. These molecules absorb light at one wavelength and emit it at a longer one. By recording only the emitted light, the microscope produces an image in which the targets glow against a dark background, revealing where they are and, under controlled conditions, how abundant they are. Practical applications include tracking processes in living cells, comparing drug responses across thousands of compounds, and, with special techniques, resolving structures below the diffraction limit.
Fluorescence also changes the shape of the data. A dataset is rarely a single grayscale image; images can have several channels (C), optical sections in depth (Z), and time points (T), most commonly stored as 12- to 16-bit values.
The optics of 2.8 still apply: the PSF, the λ/NA resolution, and the Nyquist pixel size. As a reference throughout, we use a 60×/1.4 oil objective and green emission at 520 nm, which gives a Rayleigh resolution of 227 nm and a Nyquist pixel size of 93 nm. The page first explains how labels emit light, how a fluorescence image forms, what limits its quality, and how computation can help. It then adds dimensions: several channels, depth, time, and thousands of experimental conditions. It ends beyond conventional fluorescence, with super-resolution and electron microscopy.
A fluorophore is a molecule that absorbs a photon, moves to an excited electronic state, loses a little of that energy to molecular vibrations, and a few nanoseconds later emits a photon carrying the rest. Because some energy is lost, the emitted photon has less energy, and therefore a longer wavelength, than the absorbed one. A fluorophore excited by blue light at 488 nm might emit green light at 510-530 nm. This difference, the Stokes shift, is what lets the microscope separate the faint emission from the far stronger excitation, using color filters.
The principle of fluorescence. (a) A fluorophore absorbs a photon (blue) and moves to the excited state S₁. It quickly loses a little energy to molecular vibrations (red), and after a few nanoseconds emits a photon (green) as it returns to the ground state $S₀$. Because some energy is lost, the emitted photon has less energy and a longer wavelength. Molecules can also enter a dark state (dashed), from which they blink or are permanently bleached. (b) Excitation and emission spectra of a GFP-like fluorophore. The Stokes shift between them allows filters (shaded) to separate the weak emission from the much stronger excitation.
A fluorophore is typically used to label a specific biological target. There are three common approaches:
The labeling is part of the measurement. A signal indicates where the label is, which is not always where the target is: antibodies can bind nonspecifically, and a fused GFP can alter protein behavior. Fluorophores also differ in brightness, in how quickly they fade under illumination (photobleaching, discussed below), and in how much their spectra overlap, which matters as soon as several are combined in one image. Labeling gives a model molecular information that morphology alone cannot, but it adds the protocol to the list of things that change the image: antibody batch, dye concentration, and fixation can vary between experiments as systematically as the biology.
In transmitted-light microscopy, the camera records light that has passed through the specimen. In fluorescence, the specimen itself is the light source, and the challenge is to collect its faint emission without the much brighter excitation light that produced it. Most fluorescence microscopes solve this with epi-illumination, in which a single objective both delivers the excitation light and collects the emission. Three optical elements separate the two. An excitation filter first selects the wavelengths that excite the label. A dichroic mirror then reflects this light toward the specimen, while letting the longer-wavelength emission pass back toward the camera. Finally, an emission filter removes any remaining excitation light.
Epi-fluorescence. The light source sits beside the light path. An excitation filter selects blue light, a dichroic mirror reflects it down through the objective, and the specimen emits green light, which passes back through the dichroic mirror and an emission filter to the camera (see 2.8, section How does a microscope form an image? for comparison with transmitted light).