Spinning-Disk vs Point-Scanning Confocal: Choosing by Speed, Photodamage and Optical Sectioning

Both instruments are confocal: both reject out-of-focus light with a physical pinhole in a plane conjugate to the focal plane. The choice between them is not about which is “better optics” — it is about which of three constraints your experiment is actually limited by. This page is a selection procedure, not an optics primer; for how confocal detection works at all, see Confocal Microscopy: Principle, Setup, and When to Use It.

The decision, stated once

Choose a spinning-disk system when your specimen is thin, brightly labelled, and you need many frames — fast dynamics, long time-lapses, or living samples where cumulative light dose is the limiting factor. Choose point-scanning (confocal laser scanning, CLSM) when your specimen is thick, dimly or densely labelled, when you need sectioning you can tune, or when you need a capability that only a steerable beam provides: spectral separation, a defined bleach or activation region, or a variable pinhole.

The failure mode that catches people is the third constraint. Speed and photodamage are the reasons a spinning disk gets bought; pinhole crosstalk is the reason it then fails on the one thick specimen the lab really needed it for. That limit is geometric and predictable, and the rest of this page shows how to predict it before you book time.

The disk geometry that sets every trade-off

A point-scanning confocal moves one diffraction-limited spot across the field with galvanometer mirrors and reads the descanned emission through one pinhole onto one detector. Everything about it — pinhole diameter, dwell time, zoom, scan region — is a software setting.

A spinning-disk head illuminates thousands of points at once through a rotating Nipkow disk and images the result on a camera. Almost nothing about it is a software setting: the pinholes are holes in a physical disk. Yokogawa, whose Confocal Scanner Unit (CSU) heads are the dominant design, publishes the geometry, and it is worth doing the arithmetic on it, because those two numbers determine the instrument’s entire operating envelope:

• Pinhole diameter: 50 µm (a 25 µm radius), with a 25 µm pinhole option available on the CSU-W1 for higher confocality.
• Pinhole spacing: approximately 253 µm centre to centre.

Two consequences follow directly, and neither is a vendor claim — both are division.

Consequence 1: the fixed pinhole is only correctly sized at high magnification

A pinhole’s optical effect depends on its size projected back into the specimen, which is the physical diameter divided by total magnification. The conventional yardstick is the Airy unit: 1 AU is the diameter of the Airy disk, 1.22 λ / NA, and roughly 1 AU is the standard balance between sectioning strength and signal throughput. So for a 50 µm disk pinhole at 520 nm emission:

Objective	Back-projected pinhole	1 AU at that NA	Effective pinhole
100× / 1.4	0.50 µm	0.45 µm	~1.1 AU — near optimal
60× / 1.4	0.83 µm	0.45 µm	~1.8 AU — sectioning already loosening
40× / 1.3	1.25 µm	0.49 µm	~2.6 AU — weak confocality
20× / 0.75	2.50 µm	0.85 µm	~3.0 AU — barely sectioning

This is the single most useful thing to know before booking a spinning disk. The instrument is engineered around a high-magnification, high-NA objective. Drop to 40× to capture a wider field and you have quietly given up most of the optical sectioning you came for — and no software setting will give it back, because there is no setting. On a point-scanning system the same change costs you nothing: you close the pinhole to 1 AU at the new NA and carry on.

This is also what the 25 µm pinhole option is for. Halving the pinhole moves 60× to roughly 0.9 AU and 40× to roughly 1.3 AU, restoring usable sectioning at lower magnification at the cost of throwing away signal. If you expect to work below 100×, that option is not a nicety.

Consequence 2: crosstalk is set by pinhole spacing, and scales with magnification

Pinhole spacing back-projects the same way: 253 µm divided by total magnification. At 100× the pinholes sample the specimen every 2.53 µm; at 60×, every 4.2 µm; at 20×, every 12.7 µm.

Now consider where out-of-focus fluorescence goes. Light emitted from a plane a few micrometres above or below focus arrives at the disk spread over a cone. If that cone is wider than the pinhole spacing, some of it does not hit the blocking surface of the disk at all — it falls into a neighbouring pinhole and is recorded as if it were in-focus signal from that neighbour’s position. That is pinhole crosstalk, and it is the mechanism that makes spinning-disk confocality degrade with depth.

Crucially, it does not degrade gracefully into blur, the way widefield does. It degrades into a rising, structured background haze that looks like real signal, which is why it is easy to miss until quantification produces implausible results. This is a specific, named failure — not general “worse image quality” — and it is the reason a spinning disk is the wrong instrument for a whole-mount embryo, a cleared tissue block, a thick organoid or a densely labelled section, however fast it is.

The two constraints pull in opposite directions, which is the crux of the instrument. Lower magnification widens the pinhole spacing in the specimen (less crosstalk) while simultaneously widening the back-projected pinhole (less confocality). You cannot fix both with the objective choice alone.

What vendors do about it — and the question to ask

Yokogawa’s CSU-W1 is explicitly positioned against this limit: the company states it has significantly reduced pinhole crosstalk and enables “clear observation much deeper into thick samples,” alongside an effective field of view it specifies as 17 × 16 mm — four times wider than the previous model. Note what that costs: Yokogawa specifies the CSU-W1 disk at 1,500–4,000 rpm, max 200 fps, whereas for the CSU-X1 it claims scan rates up to 2,000 fps full-frame. The head built for thick samples and wide fields is an order of magnitude slower than the head built for speed. That is the trade being made, stated in the manufacturer’s own numbers.

So when a vendor or core facility offers you “a 25 µm pinhole option,” ask precisely which parameter changed. A smaller pinhole on the same lattice improves confocality and rejects somewhat more stray light per pinhole. A wider pinhole spacing is what actually attacks crosstalk. They are different fixes for different problems, and only one of them helps with a thick specimen.

Speed: why the disk is faster, and the conditions under which it is not

The disk’s advantage is parallelism, not a faster mechanism: it excites and reads thousands of points per instant instead of one. But quoted disk figures — the 200 fps and 2,000 fps above — are disk scan rates, the rate at which the optics can deliver complete confocal frames. They are not frame rates you will achieve. Three other things bind first:

• The camera. Your actual frame rate is the sensor’s readout rate for the region you are reading, not the disk’s. A full-chip sCMOS read is typically in the tens to low hundreds of frames per second; cropping the sensor buys back speed.
• The photon budget. This is usually the real limit. Exposure must be long enough to collect a usable number of photons per pixel from your actual fluorophore at your actual expression level. A dim sample does not get faster because the disk spins faster; it just gets noisier. Speed claims are only meaningful for a specified brightness.
• Disk-rotation synchronisation. Exposure should span a whole number of disk scan periods, or the image carries fixed-pattern striping from partial scans. At very short exposures this becomes a real constraint on which exposure values are usable.

On the other side, point-scanning is not stuck at the slow frame rates it is remembered for. A resonant scanner replaces the slow galvanometer on the fast axis with one oscillating at a fixed high frequency — commonly around 8 kHz — which puts a 512-line frame in the region of 30 frames per second, and considerably higher with fewer lines. The per-pixel dwell time collapses correspondingly, so signal per pixel falls and averaging is usually needed; but the honest statement is that a resonant point-scanner and a spinning disk are in the same speed regime for many live-cell experiments, and the disk’s advantage narrows to the cases where you genuinely cannot afford the averaging.

The generalisable rule: the disk wins on speed by a large margin when signal is plentiful, and by a shrinking margin as the sample gets dimmer. Any single frame-rate number quoted without a specimen attached to it is marketing.

Photodamage: the light-dose argument and what it assumes

The standard claim is that spinning disk is gentler. The mechanism is sound: to build one frame, a point-scanner concentrates the entire excitation dose into one spot at a time, so each illuminated point experiences a very high instantaneous intensity for a very short dwell. A disk spreads the same nominal frame dose across thousands of simultaneous points, so each point sees a much lower peak intensity for correspondingly longer.

This matters because several of the processes that bleach fluorophores and damage cells are non-linear in intensity — they proceed through excited-state and multi-photon absorption pathways that scale faster than linearly, so the same total number of photons does measurably more harm when delivered in short, intense bursts. That is a real physical asymmetry, and it is why the spinning disk’s reputation for gentleness is deserved rather than folklore.

But it holds under conditions, and the conditions are frequently violated:

• It compares equal total dose. A disk that needs longer exposures because the camera is less sensitive than a good point detector may deliver more total photons per frame, eroding the advantage.
• It is per frame, and you will take more frames. The reason to buy speed is to acquire more timepoints. A gentler frame acquired ten times as often is not a gentler experiment. Cumulative dose over the whole time course is the quantity that determines whether your cells still divide normally at the end.
• Point-scanning can restrict where the dose goes. Scanning only a small region of interest, or using a longer interval with fewer averages, cuts total dose in ways a disk illuminating the whole field cannot.
• Depth changes the comparison entirely. Beyond the crosstalk regime, a disk must be driven harder to extract usable contrast from depth — which is exactly where its light-dose advantage is spent.

Photodamage is also the least transferable of the three constraints: it depends on cell type, fluorophore, medium composition, and how much phototoxic reactive oxygen your particular preparation tolerates. Treat published comparisons as directional and run your own viability control on your own sample, imaging a matched field at your intended settings and confirming the cells behave normally afterwards.

What only a steerable beam can do

Independent of speed and dose, some experiments simply require a point-scanner because they require pointing the beam:

• A continuously variable pinhole, set in Airy units at the NA in use — the ability to trade sectioning against signal per experiment rather than per purchase order.
• Spectral detection and linear unmixing, which resolves fluorophores whose emission spectra overlap too heavily for filter-based separation. A disk head uses filters.
• Defined-region photomanipulation — FRAP, photoactivation, photoconversion, optogenetic stimulation. These need the beam parked on a chosen region at chosen power. Spinning-disk systems can do this only with a separate add-on illumination path.
• Optical zoom and arbitrary scan geometry — sampling a small region at Nyquist without changing objectives, or scanning a line or a freehand shape for high-rate kinetics.
• Depth, especially with non-descanned detection or in the two-photon configuration, where excitation is confined by the physics rather than by a pinhole and no crosstalk mechanism exists at all.

Conversely, spinning disk is the better instrument when you need a large number of usable frames from a bright, thin specimen — high-content and multi-well screening, cytoskeletal and vesicle dynamics, cell-division time courses, monolayer culture, and cleared-view assays where camera-based detection with its high quantum efficiency and true parallel readout is genuinely the right architecture.

A selection procedure

1. Measure or estimate the depth you must image. Not the specimen’s total thickness — the depth of the deepest structure you must quantify. This is the first gate, because it can eliminate spinning disk outright regardless of every other consideration.
2. Fix the objective the science requires, then compute the back-projected pinhole against 1 AU at that NA using the table above. If the answer is much beyond ~1.5 AU on the disk you have access to, either secure the smaller-pinhole disk or accept that you are running something closer to a fast widefield instrument.
3. State the required temporal resolution as an interval, not “fast.” If the interval is longer than a second or two, a resonant point-scanner very likely meets it, and the decision collapses back onto sectioning and capability.
4. Check the photon budget before the frame rate. Image one field on each candidate instrument at the exposure your interval allows and look at the actual signal-to-noise. A speed spec you cannot reach because the sample is dim is not a speed advantage.
5. Check for a capability gate. Any FRAP, photoactivation, spectral unmixing or variable-pinhole requirement decides the question on its own.
6. Run a viability control at the full intended time course, not a single frame. Cumulative dose is what damages the sample.

Where both instruments clear every gate, the deciding factor is usually access rather than optics — which system your core facility supports, what the queue looks like, and how the two are costed. See Confocal Microscope Cost for how these system types are priced and what drives the difference, and Microscope Service Contract for the ongoing cost, which differs meaningfully between a laser-heavy scanning system and a disk head with a rotating mechanical assembly.

Questions worth asking a core facility before you book

• Which disk pinhole diameter is installed, and is an alternative disk available? (Determines your usable magnification range.)
• What is the deepest specimen anyone has quantified successfully on this system, and were they able to show background did not rise with depth?
• Which camera, and what frame rate does it deliver at the region size I need — as opposed to the disk’s rated scan rate?
• Is there a photomanipulation path, or is FRAP simply unavailable here?
• On the point-scanning system: is there a resonant scanner, and are GaAsP or other high-sensitivity detectors fitted? Both materially change the speed and dose comparison.

Frequently asked questions

Is spinning disk always faster than point-scanning confocal?

No. It is faster than a conventional galvanometer scanner by a wide margin, but a resonant point-scanner running around 8 kHz on the fast axis reaches roughly 30 frames per second at 512 lines and more at fewer lines, which places the two in the same regime for many live-cell experiments. And in dim samples neither instrument is limited by its scanning mechanism — both are limited by how long you must expose to collect enough photons.

How thick a specimen can a spinning disk actually handle?

There is no universal number, and any source quoting one without conditions is overreaching. The limit is set by pinhole spacing back-projected through your objective (253 µm divided by total magnification for a standard Yokogawa disk), by how much out-of-focus fluorescence your labelling density generates, and by how much the tissue scatters. A sparsely labelled, well-matched preparation performs far better at the same thickness than a densely labelled scattering one. The practical test is empirical and takes ten minutes: acquire a z-stack through the full depth and check whether background between labelled structures rises with depth. If it does, you are in the crosstalk regime and your deep timepoints are not quantitative.

Why do my spinning-disk images look worse at 40× than at 100×?

Because the pinhole is a fixed physical size and its optical effect scales with magnification. A 50 µm pinhole is about 1.1 AU at 100×/1.4 but about 2.6 AU at 40×/1.3 — the sectioning is genuinely much weaker, and this is expected instrument behaviour rather than a fault or a misalignment. On a point-scanning system you would simply close the pinhole back to 1 AU.

Can I do FRAP on a spinning-disk system?

Not with the disk itself, which illuminates the whole field through the disk pattern and cannot confine a bleach to a chosen region. It requires a separate photomanipulation illumination path fitted to the stand. If FRAP, photoactivation or optogenetic stimulation is central to the project, that is a capability gate and it points to a point-scanning system unless the disk system demonstrably has the add-on.

Does spinning disk always cause less photodamage?

Per frame, at equal total dose, usually yes — the mechanism (lower peak intensity, and damage pathways that scale non-linearly with intensity) is real. Per experiment, often no, because the reason to use the disk is to take many more frames, and cumulative dose is what determines whether the sample survives. Compare total dose across the whole intended time course, not frame to frame.

Is a spinning disk a “real” confocal?

Yes — it rejects out-of-focus light through physical pinholes in a conjugate plane, which is the defining mechanism. The qualifications are that its pinholes are fixed rather than variable, and that its rejection is incomplete in the specific, geometric way described above. Both are engineering trade-offs within confocal, not a departure from it.

Related pages

• Confocal Microscopy: Principle, Setup, and When to Use It — the underlying mechanism, acquisition parameters and artefacts common to both configurations.
• Confocal vs Fluorescence Microscope — the prior decision: whether you need confocal sectioning at all.
• Immunofluorescence: Protocol, Controls, and Troubleshooting — labelling density and background, which set how much out-of-focus light either system must reject.
• Köhler Illumination: A Step-by-Step Alignment Procedure — the transmitted-light alignment underneath every fluorescence stand, and the first thing to check when a shared instrument looks wrong.
• Phase Contrast Microscopy: Setting Up and Aligning the Phase Rings — the usual label-free companion channel for live-cell work.
• Cell Culture Reference Numbers — vessel and seeding references for planning a live-imaging time course.
• Lab Equipment — the wider instrumentation cluster.

Sources and limits of this page

Disk geometry (50 µm pinhole diameter, ~253 µm pinhole spacing, 25 µm pinhole option) and the CSU-W1 and CSU-X1 performance figures quoted above are Yokogawa’s own published specifications, cross-checked against Scientific Volume Imaging’s back-projection reference, which gives 250 nm back-projected pinhole radius and 2.53 µm back-projected pinhole spacing at 100× — consistent with the arithmetic used here. The Airy-unit figures in the table are computed as 1.22 λ / NA at 520 nm emission and are therefore approximate for any specific fluorophore. Frame-rate and photodamage comparisons are stated as mechanisms and conditions rather than fixed numbers deliberately: both depend on the specimen, the labelling, the detector and the objective in use, and a single universal figure for either would be wrong more often than right.