Confocal microscopy is a fluorescence imaging technique that uses a pinhole aperture to reject out-of-focus light, producing thin, sharp optical sections through a specimen rather than the blurred composite image a conventional widefield fluorescence microscope collects. It is the workhorse imaging method for cell and developmental biology, immunohistochemistry, and any application where researchers need to resolve structure in three dimensions or separate signal from a specific focal plane inside a thick or densely labeled sample. This guide covers the operating principle, how it compares to widefield and other scanning modalities, the acquisition parameters a researcher must set and report, sample preparation, common artefacts, and the image-integrity and reproducibility practices that make confocal data defensible in publication.
The Core Principle: Why the Pinhole Matters
In a conventional widefield fluorescence microscope, the entire specimen is illuminated at once and the objective collects emitted light from every depth simultaneously — light from planes above and below the plane of focus reaches the detector along with light from the focal plane itself, blurring the image, especially in thick or densely labeled samples.
A confocal microscope illuminates a single point (or, in spinning-disk systems, many points at once) and places a pinhole aperture in a plane conjugate to the focal plane, directly in front of the detector. Only light originating from the focal plane converges through the pinhole; out-of-focus light from other depths is defocused at the pinhole plane and largely blocked. The practical result is optical sectioning: a thin, in-focus slice of the specimen at each imaged depth, with a series of these slices (a z-stack) reconstructable into a 3D volume. This is the single feature that distinguishes confocal from widefield and explains most of its cost, complexity, and appropriate use cases.
Confocal vs. Widefield: When the Extra Cost Is Justified
Confocal is not a strictly better replacement for widefield fluorescence — it trades speed, simplicity, and gentleness on the sample for optical sectioning and improved contrast in thick specimens. The choice should follow the biological question, not default to whichever instrument is available.
Widefield is the better choice when the sample is thin (a monolayer of cultured cells, a thin tissue section), when speed matters more than z-resolution (fast live-cell dynamics, high-throughput screening), or when photodamage must be minimized because the full-field illumination, while lower in peak intensity per point than a scanning laser, is generally gentler over the timescales widefield experiments typically run. Widefield systems are also substantially cheaper to buy and maintain, and deconvolution software can computationally remove much of the out-of-focus blur post-acquisition for moderately thick samples.
Confocal is justified when the sample is thick and densely labeled (tissue sections, whole-mount embryos, organoids, spheroids), when true optical sectioning and 3D reconstruction are the actual experimental goal, when co-localization or precise spatial relationships between structures must be resolved without contaminating signal from adjacent planes, or when a specific single plane inside a thick sample needs to be isolated cleanly. The tradeoff is real: point-scanning confocal concentrates laser power into a small spot dwelling on each pixel, which increases photobleaching and phototoxicity relative to widefield, and scanning is inherently slower.
Types of Confocal Microscopy
Laser Scanning Confocal Microscopy (CLSM)
The standard configuration: a laser beam is scanned point-by-point across the specimen using galvanometer mirrors, and emitted fluorescence is descanned and passed through the pinhole to a single-point detector (typically a photomultiplier tube or a more sensitive solid-state detector — see the detector section below). CLSM offers flexible pinhole size, multiple laser lines, and generally the best resolution and sectioning control, at the cost of slower frame rates and higher per-pixel light dose, which makes it more phototoxic for extended live-cell imaging than the alternatives below.
Spinning Disk Confocal Microscopy
A spinning disk (Nipkow disk) system uses a rotating disk perforated with thousands of pinholes to scan many points across the field simultaneously, then images the result on a camera (typically a CCD or sCMOS sensor) rather than a single-point detector. This parallelization gives dramatically faster frame rates and spreads the total light dose across many points rather than concentrating it point-by-point, making spinning disk the standard choice for live-cell imaging where speed and reduced phototoxicity matter — tracking fast dynamics, long imaging time courses, or samples sensitive to light exposure. The tradeoff is somewhat reduced flexibility (pinhole size and spacing are fixed by the disk) and, for thick or densely labeled samples, more pinhole crosstalk (light from one pinhole leaking into an adjacent one) than a single-point CLSM system.
Two-Photon (Multiphoton) Microscopy — A Brief Note
Two-photon microscopy is a related but distinct technique that uses near-infrared, longer-wavelength pulsed lasers and requires two photons to arrive at a fluorophore nearly simultaneously to excite it — a nonlinear process that only occurs with meaningful probability at the tightly focused point of the laser beam. Because excitation itself is confined to the focal point, no pinhole is needed for optical sectioning, and the longer-wavelength light scatters less and photobleaches/photodamages tissue outside the focal plane less than confocal illumination. This makes two-photon the preferred method for deep-tissue imaging (hundreds of microns to a few millimeters, well beyond typical confocal penetration) and for prolonged live imaging in thick, sensitive samples such as intact brain tissue. It is not simply a “better confocal” — it requires a different (and more expensive) laser system and typically trades some resolution for depth.
Key Acquisition Parameters to Set and Report
Confocal images are only interpretable and reproducible if the acquisition settings behind them are recorded. A results section or figure legend that names the instrument model without the parameters below leaves other researchers unable to judge whether an observed difference reflects biology or acquisition settings.
- Objective NA and magnification. Numerical aperture (NA), not magnification alone, sets both resolution and light-gathering efficiency; a high-NA objective is required to realize confocal’s resolution advantage and to collect enough signal at reasonable laser power.
- Pinhole size, in Airy units (AU). Pinhole diameter is conventionally expressed relative to the diffraction-limited Airy disk size (1 AU) rather than in absolute micrometers, because the “correct” physical pinhole size depends on wavelength and objective NA. Closing the pinhole below 1 AU improves optical sectioning and resolution but rapidly reduces signal, increasing the exposure/laser power needed and worsening the signal-to-noise tradeoff; opening it above 1 AU gains signal at the cost of section thickness. Report the AU value used, not just “pinhole set to X.”
- Laser lines and power. Which excitation wavelengths were used and at what power (or percentage of maximum) directly affects both signal and photobleaching/phototoxicity; this should be recorded per channel.
- Detector type and gain. Photomultiplier tubes (PMTs) are the traditional confocal detector — robust and inexpensive but comparatively low quantum efficiency. GaAsP (gallium arsenide phosphide) detectors offer substantially higher quantum efficiency, improving sensitivity for dim signals. Hybrid detectors (HyDs), which combine a photocathode with an avalanche diode, add very low noise and fast response, useful for photon-counting and low-light applications such as FRET or single-molecule work. Detector gain (and offset) settings affect the dynamic range captured and must be held constant across images intended for quantitative comparison.
- Pixel dwell time. In point-scanning CLSM, this is how long the laser spends exciting each pixel before moving to the next; longer dwell times improve signal-to-noise but increase photobleaching and total acquisition time per frame.
- Voxel size and Nyquist sampling. To resolve the finest detail the optical system is actually capable of without wasting acquisition time (and unnecessary light dose) on oversampling, pixel/voxel size should follow the Nyquist sampling criterion: the sampling interval should be roughly 2–2.5 times finer than the optical system’s resolution limit in each dimension (lateral and axial). Undersampling loses real resolution the optics could have delivered; oversampling adds phototoxicity, file size, and scan time with no resolution benefit. Many acquisition software packages calculate a Nyquist-optimal pixel size and z-step automatically from the chosen objective, wavelength, and pinhole setting.
- Z-step size. The axial (Z) step between optical sections in a stack should also follow Nyquist sampling relative to the system’s axial resolution, which is coarser than lateral resolution — a common source of undersampled z-stacks that miss fine 3D structure.
- Frame/line averaging. Averaging multiple scans of the same line or frame reduces random noise and improves apparent image quality, at the cost of proportionally more total light exposure and longer acquisition time — a direct tradeoff against photobleaching in live samples.
Fluorophores, Channels, and Live-Sample Considerations
Each fluorophore has a characteristic excitation and emission spectrum; effective multi-channel confocal imaging depends on choosing fluorophores whose spectra are separable given the laser lines and detector filters available, and on understanding where they overlap.
Bleed-through (also called crosstalk or spectral overlap) occurs when emission from one fluorophore is detected in a channel intended for another, because their emission spectra overlap. Simultaneous multi-channel acquisition is fastest but most vulnerable to bleed-through. Sequential scanning — acquiring one channel at a time, switching laser lines and filter sets between channels rather than collecting all channels simultaneously — substantially reduces bleed-through because only one fluorophore is excited at a time, at the cost of slower total acquisition (a real consideration for live imaging, where sequential scanning across channels can introduce a small time lag between channels that matters for fast-moving structures).
Autofluorescence — intrinsic fluorescence from cellular structures (e.g., NADH, flavins, lipofuscin, collagen) or from fixation reagents like glutaraldehyde — can be mistaken for genuine signal, particularly in tissue sections, and should be checked with an unlabeled control sample imaged under the same settings.
Photobleaching (irreversible loss of fluorophore signal under illumination) and phototoxicity (light-induced damage to living cells, often mediated by reactive oxygen species generated during excitation) are the central constraints on live-cell confocal imaging. Mitigation strategies include using the lowest laser power and pixel dwell time that still yield adequate signal, minimizing the number of z-planes and time points to what the experiment actually requires, using spinning disk rather than point-scanning CLSM for extended live imaging, choosing photostable fluorophores, and using oxygen-scavenging or antifade mounting/imaging media where compatible with live samples.
Sample Preparation
Confocal image quality depends heavily on preparation choices made well before the microscope is switched on.
Mounting media and refractive-index matching. Spherical aberration (see below) is minimized when the refractive index of the mounting medium matches that of the immersion medium the objective is designed for (oil, water, glycerol, or air). Mismatches between sample, mounting medium, and immersion medium introduce depth-dependent aberration that worsens with imaging depth.
Coverslip thickness (#1.5). Most high-NA objectives are corrected for a specific coverslip thickness, conventionally #1.5 (nominally 0.17 mm). Using a coverslip outside the objective’s design tolerance introduces spherical aberration that degrades resolution and signal, particularly at high NA — a frequently overlooked source of otherwise-unexplained image quality problems.
Tissue clearing for thick samples. Because confocal penetration depth is limited by scattering and absorption in tissue, thick specimens (whole organs, large tissue blocks) intended for deep confocal imaging are often optically cleared — using one of the various clearing protocols that reduce refractive-index heterogeneity in the tissue — to extend usable imaging depth before scattering degrades signal beyond usability.
Common Artefacts and How to Recognize Them
- Spherical aberration with depth. Image quality (resolution, brightness, apparent shape of structures) degrades progressively deeper into a sample, most often from refractive-index mismatch between the immersion medium, mounting medium, and sample. Recognizable as increasing blur, dimming, and z-axis elongation of point-like structures at greater depth.
- Saturation (clipping). When detector gain or laser power is set too high, the brightest pixels exceed the detector’s dynamic range and are recorded as a flat maximum value rather than the true (higher) intensity — this destroys quantitative intensity information and should be checked for and avoided (most acquisition software offers a saturation/over-exposure indicator) before acquiring images intended for intensity comparison.
- Channel crosstalk. Signal appearing in a channel where it shouldn’t be, from bleed-through (see above) or from imperfect optical alignment between channels; recognizable as implausible co-localization or signal in a channel with no corresponding label.
- Z-drift. Slow, unintended movement of the focal plane relative to the sample during a time-lapse or long acquisition, from thermal drift, mechanical settling, or stage instability; recognizable as a gradual loss of focus over the course of an acquisition. Hardware or software autofocus/focus-lock systems mitigate this where available.
Quantification and Image Integrity
Confocal images are frequently used for quantitative claims — not just illustration — and that quantitative use carries specific methodological and integrity obligations.
Colocalization analysis. Pearson’s correlation coefficient and Manders’ overlap coefficients are the two most widely used statistics for quantifying whether two labeled structures occupy the same pixels. Pearson’s measures the linear correlation between the two channels’ intensities across all pixels, independent of absolute intensity, but does not distinguish “these structures are truly colocalized” from “these two signals happen to correlate for an unrelated reason,” and is sensitive to background and to threshold choices. Manders’ coefficients (M1 and M2) report the fraction of one channel’s signal that overlaps with the other, separately in each direction, which is more interpretable biologically but is sensitive to the intensity threshold chosen to define “signal” versus background — a choice that should be justified and reported, not left implicit. Neither statistic substitutes for adequate resolution: two structures separated by less than the system’s resolution limit will appear colocalized regardless of their true spatial relationship, and neither coefficient establishes physical interaction, only spatial coincidence at the resolution achieved.
Identical settings for compared images. Any set of images intended to be compared quantitatively — intensity, colocalization, or otherwise — must be acquired with identical laser power, detector gain and offset, pinhole size, objective, and pixel dwell time. Changing any of these between a control and an experimental image invalidates a direct intensity comparison even if both images look “correct” individually.
Permissible adjustments versus manipulation. The generally accepted standard, consistent with journal image-integrity policies, is that linear adjustments (brightness, contrast, gamma) applied uniformly to an entire image are acceptable and should be disclosed in the methods or figure legend, while any adjustment applied selectively to only part of an image, any adjustment that adds, moves, removes, or duplicates features, or any nonlinear adjustment that changes the relationship between pixels in a way that misrepresents the underlying data is not. CASRAI’s Image manipulation entry covers this distinction, including the categories most journals check for at submission, in more depth. Retaining unprocessed original files for every published image is standard practice precisely because it is the only way to demonstrate, after the fact, that only permissible adjustments were made.
Journal image-integrity policies. Most major journals and publishers now require disclosure of image acquisition and processing parameters, prohibit the manipulations described above, and in a growing number of cases run submitted figures through automated image-integrity screening before or during peer review. Authors should expect to be asked for raw, unprocessed image files on request, and should retain them regardless of whether a given journal asks at submission.
Reporting and Reproducibility
A confocal image is only as reproducible as its accompanying metadata. Best practice is to record and retain, alongside every acquired image: the full parameter set described above (objective, NA, pinhole in AU, laser lines and power, detector type and gain, pixel dwell time, voxel size, z-step, averaging), the fluorophores and antibodies used with catalog/clone information, and the mounting/preparation method. Many confocal acquisition systems embed some of this metadata automatically in proprietary file formats, but proprietary formats are not always readable outside the vendor’s own software years later, which is one of the reasons the field has moved toward open, standardized formats.
OME-TIFF, maintained by the Open Microscopy Environment (OME) consortium, is a widely adopted open file format that embeds standardized, machine-readable metadata (acquisition parameters, channel information, physical pixel dimensions) directly alongside the pixel data in a TIFF-based container, making images interpretable and reusable outside the originating vendor’s software. For public deposition, the BioImage Archive, operated by EMBL-EBI, is a general-purpose repository for biological imaging data, giving imaging datasets the same kind of persistent, citable, publicly accessible home that sequence and structure data have long had in other EMBL-EBI repositories — increasingly expected or required by funders and journals for datasets underlying published imaging figures, consistent with the same data-stewardship logic covered in CASRAI’s Data Management Plan (DMP) entry.
Frequently Asked Questions
What is the main advantage of confocal microscopy over widefield fluorescence microscopy?
Optical sectioning: a pinhole rejects out-of-focus light so each image represents a thin plane of the specimen, enabling clean 3D reconstruction and much better contrast in thick or densely labeled samples than widefield can achieve without computational deconvolution.
Is spinning disk or laser scanning confocal better for live-cell imaging?
Spinning disk is generally preferred for live-cell imaging because parallelized scanning across many points gives faster frame rates and lower cumulative light dose per pixel than point-scanning laser scanning confocal microscopy (CLSM), reducing photobleaching and phototoxicity over an extended time course, though CLSM retains an edge in flexibility and, in thick or densely labeled samples, sectioning quality.
What does “Airy units” mean for pinhole size?
It expresses pinhole diameter relative to the diffraction-limited Airy disk size (1 AU) rather than in absolute micrometers, since the physically “correct” pinhole size depends on wavelength and objective NA; 1 AU is the conventional balance point between optical sectioning/resolution and signal throughput.
Why does Nyquist sampling matter for confocal image acquisition?
It sets the pixel/voxel size (roughly 2–2.5x finer than the optical resolution limit) needed to fully capture the detail the optical system can actually resolve. Undersampling throws away achievable resolution; oversampling adds unnecessary light dose, phototoxicity, and file size without improving resolution.
Can Pearson’s or Manders’ coefficients prove two proteins physically interact?
No. Both measure spatial coincidence of signal at the resolution achieved, not physical interaction, and structures separated by less than the system’s resolution limit will appear colocalized regardless of whether they actually interact. They are evidence consistent with interaction, not proof of it, and results depend on threshold and acquisition-setting choices that should be reported.
What image adjustments are acceptable for a published confocal figure?
Linear adjustments (brightness, contrast, gamma) applied uniformly across the whole image, disclosed in the methods or legend, are generally acceptable. Adjustments applied to only part of an image, nonlinear adjustments that misrepresent relative intensities, and any addition, removal, or duplication of features are not. See CASRAI’s Image manipulation entry for the fuller taxonomy journals check against.
