Super Resolution Micros

Super-resolution
Microscopy
Essential
Knowledge
Briefings Second Edition, 2018
2 Super-resolution Microscopy
Cover image: microtubules labeled with Alexa-647. The full

width of the image is approximately 2µm. Sample and raw
images courtesy of S Niwa and N Hirokawa, University of
Tokyo, Japan. The original images were taken with a ZEISS
Elyra PS.1. The image portion in the inset shows the image
quality as obtained from conventional microscopy methods.
© 2018 John Wiley & Sons Ltd, The Atrium, Southern Gate,
Chichester, West Sussex PO19 8SQ, UK
Microscopy EKB Series Editor: Dr Julian Heath
Spectroscopy & Separations EKB Series Editor: Arne Kusserow
Super-resolution Microscopy 3
CONTENTS
4 INTRODUCTION
6 HISTORY AND BACKGROUND
10 IN PRACTICE
31 PROBLEMS AND SOLUTIONS
36 WHAT’S NEXT?
About Essential Knowledge Briefings

Essential Knowledge Briefings, published by John Wiley &
Sons, comprise a series of short guides to the latest techniques,
applications and equipment used in analytical science. Revised
and updated annually, EKBs are an essential resource for scien-
tists working in both academia and industry looking to update
their understanding of key developments within each specialty.
Free to download in a range of electronic formats, the EKB
range is available at www.essentialknowledgebriefings.com
INTRODUCTION
Super-resolution microscopy in optical microscopy
encompasses techniques that allow acquisition of images with
a resolution better than the limit imposed by the diffraction of
light (typically 10–150nm). These techniques are mainly
based on, but not restricted to, fluorescence microscopy.
Optical microscopes are the tools of choice when it
comes to visualizing the structures and dynamic processes
inside cells or within virtually any specimen. They have been
essential in life sciences (particularly biology and biomedi-
cine) from the characterization of tissues and micro-organisms
to drug development, and the diagnosis and treatment of dis-
eases. The aforementioned limited resolution of these methods
– to around 240nm in the focal plane (xy) and around 600nm
along the optical axis (z) – make individual details, such as
subcellular structures (eg membranes, vesicles and organ-
elles), appear blurred out, lacking finer detail. Compare this
with, for example, electron microscopy where resolution of a
few nanometers is commonplace, but substantial sample
preparation is required (fixation, freezing, metal coating, etc)
and extreme imaging conditions (high vacuum, bombard-
ment by charged particles, etc).
Circumventing the resolution limit while keeping the
minimally invasive nature of optical microscopy: this is what
super-resolution microscopy offers. Observing and charac-
terizing life in its finest structures, down to resolving molec-
ular details. Cellular mechanisms become quantifiable, for
example the reorganizing of actin and microtubulin fila-
ments, or the firing at individual synapses in a live neuron.
Super-resolution techniques can uncover mechanistic details

in biological processes and will therefore play a major role in
life sciences.
At the time of writing, the scientific community is at a
point where various super-resolution techniques have been
available for several years. The methods have left the field
of specialized research, in which every piece of equipment
required high levels of maintenance, and have rapidly reached
the mainstream of scientific research. It is thanks to compa-
nies like ZEISS and other manufacturers, that the scientific
community now can routinely employ super-resolution tech-
niques to answer their questions.
This Essential Knowledge Briefing provides a general
overview of the field of super-resolution microscopy. It will
explain in a simplified way how the various techniques work
and give examples of how scientists have successfully employed
these techniques in their research.
HISTORY AND BACKGROUND

The driving force in the development of optical micros
copy techniques lies in keeping its minimally invasive nature,
while removing the limitations in optical resolution.
Advances in optical microscopy, historically, focused
on improving contrast and resolution with illumination/
detection schemes such as differential interference contrast
(DIC), phase contrast, etc. A strong push forward came with
fluorescence microscopy, digital imaging and the possibility
of specific labeling (eg genetically, using fluorescent proteins
such as green fluorescent protein [GFP] or using antibodies
in immunolabeling). Fluorescence provides high experi
mental contrast – as only the structures that are of interest
become visible.
The combination of fluorescence techniques with confocal
laser scanning microscopy (LSM), in the mid-1980s, proved
extremely fruitful for biological and biomedical research. It
allowed scientists to characterize specimens with the benefit of
fluorescence specificity at the diffraction limit, with an in-plane
(lateral, xy) resolution of about 240nm and vertical (axial, z)
resolution of around 600nm (see numerous reviews and tutori-
als on confocal microscopy).
Super-resolution techniques emerged as a logical exten-
sion in this process. For a more thorough historical context, see
Cremer et al, 2013. (All references are listed in the Further
Information section at the end of this EKB.) The four approaches
in super-resolution microscopy presented here are the ones
that emerged as robust and commercially viable, all of which
are based on fluorescence microscopy:
Widefield-based approaches:
• Super-resolution through structured illumination (SR-SIM)
• Localization microscopy (direct stochastic optical recon-
struction microscopy [dSTORM], photoactivated localiza-
tion microscopy [PALM], point accumulation for imaging
in nanoscale topography [PAINT], etc)
Point-scanning approaches:
• Confocal based Airyscan
• Stimulated emission depletion (STED)
Figure 1 gives a schematic overview. The techniques, their advan

tages, limitations and practical considerations, are given in the
next section (In Practice).
There is no final statement asserting that one technique is
superior to another. Performance and usability depend strongly on
the question at hand and on the specimen that is studied. The most
important points will be laid out in the following sections in order
to allow for a more balanced comparison. Nevertheless, one will
often find a comparison in terms of resolution, as shown in Figure
2. The point spread functions (PSF) of the individual techniques,
which describe the apparent size of an infinitely small point of
light, are placed side-by-side showing the middle section along the
optical axis. The lateral (r) and axial (z) diameters of the ellipsoids
correspond roughly to the best obtainable resolution. This is often
interpreted as ‘the smaller the better’. While the most significant
resolution improvement is obtained with 3D-STED and local
ization microscopy, these techniques have some limitations in
sample preparation. On the other hand, Airyscan and SR-SIM
Widefield-based approaches Point-scanning approaches
(A) SR-SIM (B) PALM / dSTORM (C) Airyscan (D) STED
Excitation PSF
+
Depletion PSF
=
STED PSF
15–25 images 1000+ images
(rotations and phases) 10 000+ localizations
GaAsP array detector
Reconstruction in Image rendered from PSF reassignment PSF shaping
Fourier space positions Deconvolution Deconvolution
Figure 1. Overview schematic of the super-resolution techniques. The image acquisition

principle, processing details, and the final super-resolution image are arranged from top
to bottom (object of interest highlighted green – note the difference in levels of detail).
From left to right: (A) SR-SIM. Structured illumination sequences (orientations and
phases) yield images with resolution down to 100nm (xy) and 250nm (z). (B) Localization
microscopy (eg PALM, dSTORM) identifies and localizes individual molecules yielding
a final image with typical resolution of 20nm (xy) and 60nm (z). (C) Airyscan combines
raster scanning with an array detector. Reconstruction yields resolution down to 120nm
(xy) and 350nm (z). (D) STED combines raster scanning with a combination of imaging
and depletion beam. The opening of the central depletion beam determines the resolution
of the final image (typically 60nm [xy] and 100nm [z])
z
600nm
560nm
350nm
250nm
~100nm
~60nm
r
240nm 70nm 120nm 100nm ~80nm v~24nm
LSM STED Airyscan SR-SIM 3D-STED 3D-PALM
Figure 2. PSF of the various techniques represented by ellipsoids with a lateral (r) and
axial (z) diameter equivalent to the most typically quoted resolution. Drawn to scale. The
individual advantages and drawbacks of the techniques are not represented in this graph.
Numbers were taken from Schermelleh et al, 2010 and may not represent individual
manufacturer’s specifications
are broadly employable and deliver high resolution – albeit not as

high as the one obtainable with localization microscopy.
It is important to emphasize that the quality of the samples
will affect the attainable resolution and that the values reached in
practice, for example from a noise-limited biological sample,
may be significantly (10–20%) larger than the numbers given in
Figure 2. The resolution of a technique or of an instrument is
often characterized using very clean and bright technical samples
(patterns, beads, etc). As a recommendation for choosing the
technique that would be most suitable for a given study or speci-
men, the researcher should consider many aspects besides the
resolution – such as sample preparation and temporal resolution.
IN PRACTICE
Super-resolution with structured illumination microscopy
SR-SIM combines fluorescence, widefield-based structured
illumination and digital image reconstruction. The illumina-
tion is structured using a sequence of known grating patterns,
leading to the reconstruction of the image with up to two-fold
improved resolution (or 120nm in xy and 300nm in z). SR-
SIM emerged in the 1990s, developed by, amongst others,
Mats Gustafsson, at the University of California San Fran
cisco, USA.
The raw images in structured illumination (Figure 3, top
row) show the grid patterns (constant and consistent over the
entire image) that are ideally adapted to the optics of the
microscope (numerical aperture, laser wavelength, etc). Acqui
sition and reconstruction (left panels of Figure 3, middle and
bottom rows) are automated on modern super-resolution
microscopes (matching of grating to optics, adaptive recon-
struction filters, etc).
SR-SIM uses the information contained in the known illu-
mination structure (ie its frequency and angle). The resulting
patterned images are a product of the grating with the structure
of interest (eg the tubulin mesh in a cell). When grids of fine
lines overlap at an angle, they create distinct patterns of coarser
lines running across the image (or moiré, eg as seen in daily life
through folded curtains). The combined image with the coarse
pattern can be separated into individual components (in the
frequency domain) and the known illumination pattern can be
accounted for. Small objects (high frequencies) contribute to
the coarse patterns (ie they have been down-transformed),
Figure 3. SR-SIM acquisition process on Alexa-561 labeled tubulin structures. Top row:
shows a subset of raw images with differently shifted and oriented grating patterns.
Middle row: shows the reconstructed super-resolution image on the left and, as a
comparison, the same area imaged with confocal microscopy. Bottom row: shows
zoomed-in views of the indicated orange squares
while large objects (low frequencies) remain unaffected. Once

separated and cleaned from the grating, the frequencies can then
be transferred back to their original positions, and this is then
regenerated as the proper image of the small object. The net result
is an image with resolution improved by up to a factor of two.
The grating separation is at the heart of SR-SIM: accurate
separation leads to high-quality images. The resolution also
depends on the angle (orientation) and phase (position) of the
grating: this is why the grating appears to be swept across the
image at different angles and phases. A typical single image in
SR-SIM is reconstructed from between 15 and 25 individual
images (three or five angles with five phases each).
Main advantages and limitations of SR-SIM

Versatility and live-cell imaging: SR-SIM is compatible with
live-cell imaging conditions and is not restricted to specific
wavelengths or laser powers.
Acquisition speed: SR-SIM can acquire images from a large
area (eg 80 x 80µm) at 1 to 10 frames per second. Simultaneous
acquisition of multiple colors is possible when using multiple
cameras and appropriate beamsplitters.
Resolution in 3D and depth: the resolution improvement (ie
100nm in xy/300nm in z) is typically obtained from images
taken up to a distance of 20µm from the coverslip surface.
Data analysis: data analysis, like co-localization or distance
measurements, is straightforward. Noise is visible as texture
and can be affected by reconstruction filter settings. Texture
can be greatly minimized by increasing the number of grating
angles (eg from three to five).
SR-SIM needs reconstruction algorithms. Reconstruc-
tion artefacts (textures) can appear as intensity fluctuations
and depend on the quality of the initial ‘raw’ data. Poor image
quality can have various origins: from instrument alignment
issues to poor sample preparation. A number of SR-SIM data
verification protocols and software packages (eg SIMCheck,
for ImageJ) can be used to help or guide the user.
Localization microscopy (PALM/dSTORM)

Localization microscopy, such as PALM and dSTORM (more
details in Blom et al, 2017), relies on the microscope’s ability
to determine the position of individual fluorescent molecules
located along a structure of interest, rather than resolving them
optically. The positions can be determined with a precision of

the order of 10nm. If thousands of such positions are gathered
and superimposed, then it is possible to generate an image of
a structure with improved resolution. The resolution depends
on the size and density of molecules and the obtainable signal-
to-noise ratio. It is therefore theoretically unlimited. Typical
images, however, provide 10-fold improved resolution in
comparison to conventional microscopy (20nm in xy and
60nm in z).
In localization microscopy, the challenge lies in detecting
molecules on a one-by-one basis in order to resolve the struc-
ture of interest (eg a tubulin fiber). In most cases, this is achieved
on the time axis, distinguishing between individual molecules
via their on/off behavior (roughly: from a dark state to an emis-
sive state – see the very simplified Jablonski–Perrin diagram,
Figure 4, top row). In a sequence of images, the individual mol-
ecules are visible as diffraction-limited patterns that appear and
disappear (on/off) as the sequence progresses (see Figure 4, top
row, right). The sequences of (typically 10 000 and more) frames
are acquired using, for example, widefield illumination or total
internal reflection approaches (TIRF) as well as sensitive, fast,
camera detection. Analysis of the sequence gives the individual
molecule’s positions, which are plotted in the final image with
improved resolution. Figure 4, bottom row, left, shows a recon-
structed image in which the individual positions of each spot
(or molecule) are superimposed.
The two most common approaches that allow observa-
tion of individual molecules via their on/off behavior are
PALM and dSTORM – the difference lies in the switching
S1
T1
D
S0
Fluorescence Dark states
Figure 4. Localization microscopy. The upper row illustrates the typical on/off behavior of
single molecules as seen in a time-series (left, a simplified Jablonski–Perrin fluorescence
diagram). Individually, molecules are visible as diffraction-limited patterns that switch on
and off in time (the last off step is usually irreversible). The graph shows a typical intensity
transient for a single molecule. The two images below show segments of microtubules, as
reconstructed from the individual molecule positions (left) with around 20nm resolution
and the corresponding widefield image (right) for comparison (sample using standard
Alexa 561 immunolabeling and embedding)
mechanism. PALM, developed by Eric Betzig and colleagues

at Howard Hughes Medical Institute in Virginia, uses photo-
activation to switch the molecules. In a sequence of images,
the activation can be set so that only a few molecules appear
in each frame of the sequence. They can therefore be easily
identified and distinguished from one another in the final
sequence. dSTORM relies on the physico-chemical interaction
of fluorescent dye molecules with their immediate surroundings,
which cause the molecules to switch on and off (hence the

term ‘stochastic’). Under proper conditions (pH value, redox
states, etc) only a few molecules are on during the acquisition
of each frame and therefore easily distinguishable from one
another in the sequence. For completeness, we would also like
to mention a third approach: PAINT. It does not rely on a direct
on/off switching of the dye molecules, but rather their appear-
ance and disappearance from the images due to binding and
unbinding. When the dyes bind they remain immobilized and
are visible as spots that can be localized; when they unbind
they diffuse rapidly and therefore remain undetected).
For the final reconstruction step – ie the recognition of
individual molecules and mapping them out on a final image
– it is of little relevance whether they were photoactivated
(PALM), if they were blinking passively (dSTORM), or if they
were only appearing and disappearing from the images
through some other process (PAINT). There is some differ-
ence, however, between the two switching approaches from
the experimental side.
In dSTORM, label molecules emit at random times due
to chemical reactions or interactions in their immediate vicin
ity (eg cis-trans-isomerism, complexation with reactive oxy-
gen species, etc). There is little exterior influence or control
over the experiment except for adding chemical reagents that
will have to reach the immediate vicinity of the dye molecule
(this is also not trivial if one considers compartmentalization
or hydrophobicity, for example). On the positive side, the
organic dye molecules are usually very bright and are more
adaptable to the experimental circumstances (eg using far-red
dyes, such as Cy5, that emit outside of the autofluorescence

spectrum).
PALM employs photoactivatable dyes (predominantly
switchable fluorescent proteins, such as photoswitchable
GFP, tdEOS, etc). The switching of the individual molecules
is still random, but the rate with which the molecules switch
on or off can be controlled by increasing or decreasing the
intensity of the switching laser (eg 405nm). Fluorescent pro-
teins are around 5nm in size and are genetically encoded into
the structure of interest (they can be used in vivo, have a
higher specificity and do not require fixation and permeabi-
lization of the specimen). On the negative side, fluorescent
proteins can exhibit maturation issues; they can disturb the
expression levels of the protein of interest and are compara-
tively dim.
An important question in localization microscopy is:
‘how many molecules are needed to get a good image?’ The
answer: it depends on the size of the structure itself. Roughly,
one would need (at least) one fluorescent molecule in every
20nm. This is comparable to the size of a large protein. Thus,
obtaining the required labeling density is not a trivial task. In
the same vein, the labeling molecules themselves (including
functional groups or primary and secondary antibodies) are
in a size regime comparable to that of the positioning accuracy.
So, while the positioning accuracy for a single fluorophore
can be smaller than 20nm, this is of little use if the structure
was labeled only with a few molecules, and if the labeling
molecules themselves (eg primary and secondary antibodies)
were too large.
Advantages and limitations of PALM/dSTORM

Versatility and live-cell imaging: dSTORM is generally not
considered to be an adequate technique for live-cell imaging
because the samples are usually prepared using immunolabel-
ing (fixed cells). Localization microscopy, however, has been
used to image living specimens with meaningful data from
the perspective of molecule or particle tracking (Muriel et al,
2011; Nan et al, 2013).
Acquisition speed: PALM and dSTORM are considered slow
because collection of a typical image sequence (>1000 frames)
takes upward of 10 seconds, typically minutes.
Resolution in 3D and depth: in practice, PALM/dSTORM
deliver the highest resolution of all presented super-resolution
methods (theoretically unlimited – typically 20nm in xy/
60nm in z) and can deliver molecular detail. Best results are
obtained from transparent and well-prepared specimens near
(around 10µm) the coverslip surface.
Data analysis: classic data analysis (distance measurements
and 3D reconstruction) is possible. The capability to detect
individual molecules, however, opens a wealth of data analy-
sis possibilities that are not easily accessible via other meth-
ods (cluster analysis, intermolecular distances, etc). PALM/
dSTORM images are rendered from a table of localized mol
ecules (accurate channel alignment and drift compensation
needs to be in place). Quantitation requires careful calibra-
tion and controls.
The biggest challenge for PALM/dSTORM is the need for
photoswitchable molecules or addition of chemistry to bring
the labels into an adequate ‘blinking’ regime. Also, PALM and
dSTORM have limited in vivo applications. Long-term stabil-

ity is a crucial concern for PALM/dSTORM equipment.
Airyscan
Airyscanning is a super-resolution approach related to confo-
cal LSM. To briefly recap, before moving on to Airyscanning,
conventional LSM raster-scans a focused excitation beam over
the specimen. Fluorescence originating at that focal spot is then
separated from off-focus fluorescence with a pinhole. This
rejection and the accompanying increase in resolution depends
on the pinhole size. A small pinhole increases the rejection, and
therefore, increases the resolution (like the tip of a sharp pencil
in comparison to a larger, blunt tip). When the pinhole size
matches the illumination spot diameter (1 Airy unit, AU) most
signal is collected at optimal rejection of the off-focus light. The
resolution of conventional LSMs is given for a 1AU opening of
the pinhole, about 240nm in xy, and around 600nm in z.
The pinhole opening can be made arbitrarily small, eg to
a fifth of the illumination spot diameter (0.2AU). It turns out
that the accompanying resolution improvement will be limited
(depending on the wavelength and optics used) to a factor 1.4
enhancement (to around 180nm) in comparison to the classi-
cal diffraction limit. More important, however, is the strong
rejection of signal. At 0.2AU the rejection is 20-fold (95%),
and this proves prohibitively impractical for biological sam-
ples (and even more so for live specimens).
The key element in Airyscanning is the detector design,
which acts as an array of closed pinholes, each providing the
aforementioned resolution increase, but collectively not
0.2AU 1.25AU
Figure 5. The top row shows the Airyscan detector array. It consists of individual
detectors that correspond to a pinhole closed to 0.2AU. The individual images are offset
and need to be rearranged (right panel, top row, shows the images from two individual
detectors of the array). The final result is shown in the bottom row images. Left: the
reconstructed Airyscan image. Right: confocal image for comparison. Microtubules
labeled with Alexa 561 (as in the previous figures)
suffering from a high rejection of signal. This is because the

detectors are adjacent to each other: the light rejected by one
detector subunit is collected by its neighbors. This simple and
elegant idea was formulated early in the development of LSMs
(Cox et al, 1982), and revisited experimentally more recently
(De Luca et al, 2013). A fast enough, sensitive and low-noise
detector array became commercially available in 2014.
Airyscan addresses the challenge of imaging biologically
relevant samples (low expression levels and keeping low laser
dosage) without signal rejection at closed pinhole. It is a new

detection standard for LSM imaging, allowing entry into the
realm of super-resolution. It is compatible with any form of
LSM, including two-photon microscopy, adding tremendously
to the versatility and applicability of the approach.
The data collected with an Airyscan detector array
requires pixel reassignment (deconvolution, a computational
approach broadly employed in LSM), which is a simple com-
putational step. In total, the resolution improvement can
reach 120nm laterally (xy) and 350nm axially (z).
Advantages and limitations of Airyscan

Versatility and live-cell imaging: of all presented super-
resolution techniques, it is the most robust under live-cell
imaging conditions. Airyscan is not restricted to specific
wavelengths, laser powers or special objective lenses.
Acquisition speed: depends on the scanned area. Small areas
(eg 10 x 10µm) allow speeds upward of 30 frames per second,
large areas (eg 60 x 60µm) require a few seconds per image.
Resolution in 3D and depth: the 3D resolution improvement
is approximately two-fold in comparison to LSM (eg 120nm in
xy/350nm in z). Best quality images are readily obtained from
transparent and well-prepared specimens. However, Airyscan
can also be used in conjunction with large working-distance
objectives and two-photon excitation conditions, and can be
used to reach deep (millimeters) into tissue preparations.
Data analysis: the reconstruction of the images is straightfor-
ward and fast on modern computers. Airyscan, just like con-
focal LSM, allows for quantitative data analysis (eg imaging
and quantifying calcium bursts, photorecovery in diffusion,

co-localization).
Of all presented super-resolution methods, Airyscan
provides a resolution improvement comparable to that of SR-
SIM. As with all raster scanning methods, the acquisition
speed depends on the scanned area and the required pixel size
(assuming a fixed dwell time per pixel). However, since it is an
LSM-based approach it is very flexible (eg zoomed-in scan-
ning, photomanipulation, combination with two-photon
imaging) and less sensitive towards changes in refractive
index and inhomogeneity in the specimen.
Stimulated emission depletion microscopy

STED microscopy is related to confocal LSM in the sense that
it, too, is a point-scanning method. However, instead of using a
pinhole to reject off-focus light (see the previous section on
Airyscan), it suppresses off-focus emission (via stimulated
emission) right at the illumination spot on the sample. To
accomplish that, a second, high-intensity donut-shaped deple
tion beam is superposed with the focused imaging beam. The
size of the ‘opening’ of the depletion donut therefore deter-
mines the obtainable resolution – the smaller this opening, the
higher the resolution (as, in analogy, the sharper the pencil, the
smaller the tip). This elegant approach was pioneered in the
mid-1990s by Stefan W Hell’s group, now at the Max Planck
Institute for Biophysical Chemistry in Göttingen, Germany
(see Eggeling et al, 2015). A particularly interesting aspect of
STED is that the donut ‘opening’ could, in principle, be made
arbitrarily small. Therefore, the resolution is, in principle, not
limited (see Figure 6 for schematic illustration). The depletion

beam can be shaped in 3D in order to also enhance resolution
along the z-axis. In practice, the typical resolution given is of
the order of 50nm in xy and 80nm in z.
The image formation is analogous to the image forma-
tion in confocal LSM, ie via raster scanning the beam
across the specimen and detecting (and mapping) the flu-
orescence intensity in the final image. The key element
(and the key difference) in STED (in comparison to confo-
cal LSM) is the presence of the second beam for depletion
via stimulated emission: the aforementioned donut-shaped
beam, which is centered with the first (say, conventional)
imaging beam.
(a) (b) Phase
2π
OBJ 0 Y
PM X
DM1
STED (c) 1
Amplitude (AU)
DM2
Ex
0
D 500 550 600 650 700 750 800 850
Wavelength (nm)
Figure 6. Simplified schematics of a STED microscope (left panel). The excitation and
STED beams are superimposed by dichroic mirrors (DM1, DM2) and focused in the
sample. A helical phase mask (PM) in the STED beam path creates a donut-shaped
STED focus in the sample (red pattern, PM schematics in top right). Fluorescence is
collected by the objective and focused onto a detector (D). Images are generated via
raster scanning. Right, bottom, shows a hypothetical excitation (dotted line) and
emission spectrum of a fluorophore. Figure from: Kubitscheck U (Ed), Fluorescence
microscopy: from principles to biological applications, first edition, 2013. Copyright
Wiley-VCH Verlag GmbH & Co. Reproduced with permission
From the perspective of experimental usability, the chal-

lenges associated with STED are threefold: first, maintaining
a stable and precise shape of the donut, perfectly centered
with the imaging beam; second, conceiving the experiment so
the dyes and labels used are within the spectral and photo-
physical confines of the depletion beam (not all dyes can be
depleted efficiently); and third, adapting the depletion beam
intensity to minimize the dosage on the specimen.
To stay within the confines of the available spectral win-
dows, it is necessary to use ‘depletable’ dyes. Not all dye mole-
cules perform in the best possible manner for STED studies,
but the list is increasing at a fast pace. Coping with a very high
intensity laser (increased bleaching, increased toxicity, heating,
photodamage) brings further constraints into the experimental
planning, but the newest technology in STED (eg rescue STED)
is minimizing that problem. Commercially, STED is available,
for example, as a stand-alone system or as an add-on setup for
various microscopy stands (eg through Abberior Instruments,
Göttingen, Germany).
Advantages and limitations of STED

Versatility and live-cell imaging: STED is generally not con-
sidered the most adequate technique due to the high power
required for the depletion beam. Live-cell measurements have
been demonstrated, however.
Acquisition speed and FOV: as a point-scanning-based
method, the frame rate of STED depends on the FOV used,
and the resolution under which it is operated. For small
zoomed-in regions (10µm x 10µm, for example) speeds in
excess of 50 frames per second can be acquired. Larger FOVs

require a few seconds. Simultaneous acquisition of multiple
colors is possible.
Resolution in 3D and depth: the typical resolution is 50nm
in xy and 80nm in z, but is theoretically unlimited. Additional
image analysis (and deconvolution) allow for further resolu-
tion improvement. The depletion beam can also be shaped
along the z-axis, giving resolution in z of about 80nm (at a
slight expense of lateral resolution). STED becomes challeng-
ing when going into thick, over-labeled, noise-rich, scatter-
ing specimens.
Data analysis: all data analysis, such as co-localization, size
and distance measurements are possible with STED images.
Also, highly quantitative fluorescence correlation techniques
(FCS) have been demonstrated using STED-based setups
(Eggeling et al, 2015).
The most frequently found point of criticism of STED is
its requirement for high-power laser and the concomitant
phototoxicity and increased photobleaching rate. Most
efforts that are put into STED and STED-related technology
are aimed at keeping the depletion laser stable and as low
as possible.
Considerations on sample preparation for all methods

Super-resolution microscopy requires thorough sample prep
aration. Image quality is rapidly affected by impurities (dust
grains, bubbles, unspecific staining, etc) on the specimen being
studied – for sample preparation see, for example, Allen et al,
2013. In essence, the message is simple: care must be taken
that all parts of the system, from the coverglass to the mounting
or embedding medium, are clean and well defined (eg uniform
thickness, clean mounting, labeling specificity).
Dyes and labels play an important role in fluorescence
microscopy. Fluorescent proteins (eg GFP, mCherry, tdEOS)
and organic dyes (rhodamines, cyanines) are available in var-
ious different colors and functionalities (eg antibodies, suc-
cinimide esters) allowing scientists to visualize cellular pro-
cesses or interactions between organelles by labeling two or
three groups of proteins (or DNA, lipids, etc) with different
color labels. The labeling approach depends strongly on the
specific topic that is being studied. Also, not all super-
resolution techniques can be used in conjunction with all
available dyes. Below, we present a few remarks on the topic of
labeling, specific to each technique presented in this guide.
Airyscan and SR-SIM can work with the same fluores-
cent labels as conventional fluorescence microscopy. These
techniques impose the least restriction when it comes to
choice of fluorescent labeling.
STED requires a careful combination of fluorescent dyes
considering the used depletion laser wavelengths. The list of
dyes that have been used and tested in STED is rapidly grow-
ing, but it is, in comparison, still much smaller than the list
used for Airyscan and SR-SIM. Labels for STED are exposed
to high-intensity light and therefore need to be stable and
robust enough to withstand repeated excitation.
PALM/dSTORM require labels that can be chemically
or optically switched so that they only fluoresce for a short
time – not all dyes do that and not all recipes are guaranteed
to work (for example, if the dye molecule is not accessible to

the chemical reagents due to compartmentalization or solu-
bility issues). The list of dyes (and preparation protocols) for
PALM and dSTORM is rapidly growing.
CASE STUDY 1. SR-SIM

SR-SIM was recently employed by Professor Mariana Pinho’s
research group at the Laboratory of Bacterial Cell Biology, Insti-
tuto de Tecnologia Química e Biológica, Portugal, to reappraise
the questions surrounding bacterial cell-division. In their pub-
lication (Monteiro et al, 2015) they state that it was the small
size of bacterial cells that called for studies employing super-
resolution methods. With super-resolution they were able to draw
a clear picture of asymmetry generated at the cell division of
S. aureus bacteria.
S. aureus COL cells were labeled for five minutes with WGA-488 (peripheral cell
wall dye, green), washed and stained with Nile Red (membrane dye, red). Cells
were then placed on an agarose pad, allowed to grow at room temperature and
imaged by SR-SIM. Upon division, the old cell wall preserved the green WGA-488
signal while the new surface is labeled only in red. Scale bar 500nm. Images taken
with a ZEISS Elyra PS.1 microscope using a Plan-Apochromat 63x/1.4 Oil DIC
M27 objective. Images were acquired using five grid rotations
CASE STUDY 2. STED

STED, with its high-resolution capabilities and optical sectioning,
is well suited to image at the nanoscale in tissue or organisms. In
their publication, researchers Unnersjö-Jess and colleagues (from
KTH and the Karolinska Institutet, Sweden) show that STED
imaging is well suited to resolve the filtration slit in cleared kid-
ney tissue (Unnersjö-Jess et al, 2016). The glomerular filtration
barrier, consisting of podocyte foot processes with bridging slit
diaphragm, glomerular basement membrane and endothelium,
is a key component for renal function and tools that allow for
their 3D structural characterization in health and disease are
highly desirable. It was the combination of clearing techniques
and deconvolution applied to STED imaging that provided the
required resolution and signal quality.
Top-view of the filtration slit in podocyte foot-processes from a cleared kidney

biopsy where the two slit proteins podocin (green) and nephrin (magenta) were
immunolabeled with antibodies carrying Alexa594 and AbberiorSTAR635 and
super-resolved with a 775nm STED laser
CASE STUDY 3. PALM

The dimensions of neuronal synapses, of the order of 100nm,
suggest that optical super-resolution imaging methods are neces-
sary for thorough investigation of protein distributions. In their
publication (Liebmann et al, 2013) scientists from the Karolin-
ska Institutet in Sweden applied localization microscopy tech-
niques (PALM) to resolve synaptic protein topology and quantify
single molecules of the neuronal sodium pump. Results revealed
a compartmentalized distribution of sodium pumps in dendritic
spines, with several nanoclusters of pumps typically found in the
spine head and fewer in the spine neck.
Overview of a cultured hippocampal neuron genetically expressing the neuronal

sodium pump (a3 isoform) labeled with the fluorescent protein PA-eGFP.
The inset shows a PALM-generated pointillistic single molecule image of the
sodium-pump topology in a dendritic spine (blue square). The image was acquired
on a ZEISS Elyra PS.1 with 488nm excitation and 405nm photoactivation. Scale
bar 300nm
CASE STUDY 4. Airyscan

Most neuronal function is regulated via membrane trafficking, in
which the transmembrane proteins Nsg1 and Nsg2 play a critical
role. Nsg1 has previously been identified as helping regulate
endosomal recycling and sorting. However, in contrast to previ-
ous conclusions, Dr Bettina Winckler’s lab in the Department of
Cell Biology at the University of Virginia, USA, has demonstrated
with Airyscan technology that Nsg1 and Nsg2 proteins are not
resident endosomal proteins but traffic rapidly between the cell
surface and lysosomes (Yap et al, 2017).
Cells were fixed in 2% paraformaldehyde/3% sucrose/phosphate-buffered saline

(PBS) in 50% conditioned medium at room temperature for 30 minutes, quenched
in 10mM glycine/PBS for 10 minutes. The fixation conditions used do not introduce
holes into the overwhelming majority of cells. Coverslips were then blocked in 5%
horse serum/1% bovine serum albumin (BSA)/PBS ± 0.2% TritonX-100 or 0.1%
saponin for 20 minutes. Antibodies were diluted in 1% BSA/PBS and incubated for
one hour. The circles show endosomal structures that were picked up for analysis.
Blue channel was Nsg1, Red Nsg2 and Green LAMP1 (lysomal-associated membrane)
PROBLEMS AND SOLUTIONS

With the availability of super-resolution microscopes,
and their capability of achieving resolutions in the range
10–150nm, it is not surprising that the scientific community
has rapidly incorporated the techniques into their research.
Super-resolution techniques are now in widespread use.
Despite this success, however, and as mentioned above, all
techniques have certain constraints. In this chapter, we pres-
ent these constraints again as a side-by-side comparison, and
point (when possible) towards strategies to overcome or min-
imize certain limitations.
Versatility/live-cell imaging
Airyscan: is perfectly compatible with live-cell imaging. It
outperforms conventional confocal or two-photon LSM in
terms of gentle imaging conditions (sensitivity increased by a
factor of 4–8). There is no restriction to specific laser wave-
lengths or specific objectives. In other words, it is straight
forward to acquire four channel images across the visible
spectrum (from blue to far red) in an incubated glass-bottom
petri dish, with the cell culture in conventional (but phenol-
red free) culture medium, at 37°C using, for example, a 63x/1.2
water immersion objective.
SR-SIM: is compatible with live-cell imaging; there is no
requirement for particularly high laser power or specific dyes
to achieve super-resolution. It is straightforward to acquire
four channel images across the visible spectrum (from blue
to far red) in an incubated glass-bottom petri dish, with the
cell culture in conventional (but phenol-red free) culture
medium, at 37°C using, for example, a 63x/1.2 water immer-

sion objective.
PALM/dSTORM: is not considered compatible with live-cell
imaging. While there is no requirement for particularly high
laser power to achieve super-resolution, it is very difficult to
acquire meaningful data in terms of super-resolution images
and in the context of live imaging. Photoactivation of individ-
ual fluorescent proteins (PALM) and other types of switching
(eg passive) can be achieved in live-cell imaging, but the choices
of fluorophores are limited.
STED: brings the highest irradiation dosage to the specimen,
because of the high-intensity depletion beam. It is therefore
considered the least gentle super-resolution imaging technique.
It has been demonstrated early on in the context of live-cell
imaging (Bottanelli et al, 2016), but it still remains a challeng-
ing modality for STED microscopy. Moreover, it is restricted to
the dyes that can be depleted by the offered laser wavelengths.
Considerable effort is being put into development of STED
microscopes that utilize less power in the depletion beam
(eg using pulsed lasers and gated detection). Alternatively, one
can also deliberately choose to decrease the obtainable super-
resolution (to levels similar to those of Airyscan or SR-SIM) by
reducing the laser power of the depletion beam.
Speed/live-cell imaging
SR-SIM: delivers the highest acquisition speed at full FOV.
As a camera-based method, SR-SIM acquires a large FOV (eg
66µm with around 1200 x 1200 pixels) in a very short time
(around 1–10 frames per second – limited by the time required
to move the grating orientation and frame rate of the camera).

Simultaneous acquisition of multiple colors is possible when
using multiple cameras (eg using dual camera adaptors) and
appropriate beam splitters.
Airyscan: as a point-scanning based method, the frame rate
of Airyscan depends on the FOV used. For small zoomed-in
regions (eg 10 x 10µm) speeds in excess of 30 frames per sec-
ond can be acquired. Large fields of view, however, require a
few seconds.
STED: as a point-scanning based method, the frame rate of
STED depends on the FOV used and the number of pixels.
For small zoomed-in regions (eg 10 x 10µm) speeds in excess
of 30 frames per second can be acquired. Large fields of view
require a few seconds. Simultaneous acquisition of two colors
and FCS is possible.
PALM/dSTORM: These techniques are not considered fast.
While a large FOV (eg 66µm) is acquired in a single shot of a
few (10 or so) milliseconds duration, there is also the necessity
to acquire thousands of individual images. This brings the tem-
poral resolution of PALM/dSTORM images to the timescale of
minutes. This is sufficient temporal resolution for the obser
vation of very slow cellular processes. Fast processes, however,
on the molecular scale, need to be examined via tracking indi-
vidual molecules. While this approach does not deliver super-
resolution images per se, the information from an individual
molecule (in the form of its trajectory) brings a tremendous
amount of dynamic information. Simultaneous acquisition of
multiple colors is possible when using multiple cameras (eg
using dual camera adaptors) and appropriate beam splitters.
Depth
Airyscan: could be considered the technique of choice when it
comes to imaging in deep tissue or thick specimens. In partic-
ular, combining two-photon LSM with Airyscan detection
offers an interesting way to reach deep into thick tissue prepa-
rations or living organisms. Airyscan is less sensitive towards
changes in refractive index, scattering and inhomogeneity in
the specimen and it can perform better than all other super-
resolution techniques when used deep (tens of micrometers)
inside tissue.
SR-SIM: performance depends on the accurate (and high
contrast) projection of the illumination pattern into the spec-
imen. While this is easily achieved in transparent and
well-prepared specimens (eg cell cultures, bacteria, thin plant
roots, C. elegans, zebrafish embryos and, in general, objects
close to the coverslip surface), it rapidly becomes challenging
when going into thick, over-labeled, noise-rich, scattering
specimens; also very sparse, dim and quickly bleaching struc-
tures. Typically, best results are obtained from images taken
up to 20µm distance from the coverslip.
PALM/dSTORM: the best results are obtained from regions
close to the coverslip surface (around 10µm) and this is
achieved easily in transparent and well-prepared specimens
(eg fixed cell cultures, bacteria, thin plant roots, objects close
to the coverslip surface). While PALM/dSTORM can operate
under TIRF illumination conditions (and therefore delivers
very high contrast) this aspect will be omitted here as there is
virtually no depth information in this mode of operation. In
terms of depth, PALM/dSTORM becomes more and more
challenging when going into thick, over-labeled, noise-rich,

scattering specimens.
STED: the problems of STED with respect to depth of imaging
are twofold. First, the high-intensity depletion beam needs to
generate a well-shaped donut in order to provide the resolution
enhancement. Projecting a well-defined donut through a thick
specimen is difficult and the quality rapidly decays with pen
etration depth. Second, the high-power depletion beam has to
cross the entire specimen, so also places a substantial laser dos-
age on the image planes that are not being imaged.
WHAT’S NEXT?
Optical super-resolution methods have reached the main-
stream of scientific research thanks to the commercial avail-
ability of robust and well-designed microscopy platforms. The
main driving force for the further development and improve-
ment of these methods, however, comes from a vibrant and
active community (of networked users and experts) that con-
stantly pushes forward for newer, improved or alternative
techniques and analysis methods, benchmark tests and proto-
col collections.
The main goal is to apply the techniques to live-cell imag-
ing situations. Here, the specimen is usually, sensitive, dim
and mobile in comparison to fixed samples (bright and immo
bile). The challenge, therefore, is to attain the highest possible
super-resolution, with high contrast, acquired faster and with
less laser power, in thick scattering tissues, with minimal
labeling and online functional data processing. (At the same
time, of course, the equipment should be economically viable
and easy to use by non-experts.)
SR-SIM: the bottleneck, currently, lies in the acquisition of
15–25 raw images per final reconstructed super-resolution
image. For this reason, one of the main challenges in SR-SIM
is to produce microscope platforms that allow us to minimize
the acquisition time in a robust manner (for example by
changing to different, more time-efficient, illumination
patterns). Another challenging aspect in SR-SIM is related to
data reconstruction and data handling, which is currently an
additional post-acquisition step on most platforms. Faster
and more robust, easier-to-use reconstruction software (that
would perform a raw data quality check as it is being acquired,

for example) is highly desirable.
Airyscan: allows for fast imaging over small areas, for exam-
ple 100ms per image (10 frames per second). The speed drops
dramatically for large fields of view to a few seconds per frame.
The most promising path to address this problem is to paral-
lelize acquisition by using multiple foci or modulating the
PSF (Fast mode for ZEISS Airyscan is a step in that direction).
STED: (as a point-scanning based technique) allows for fast
scanning over small fields of view. However, the high resolu-
tion obtainable with STED requires more (smaller) pixels to
be scanned for a given FOV. This means it can take 10–30
seconds to produce a single STED image. Parallelization (using,
for example, multiple foci) seems challenging, as it would also
require multiple depletion beams. Gating and pulsed beam
approaches, different forms to shape (also in 3D) the deple-
tion beam, could improve this situation. Significant effort is
also being channeled into the development of adequate dyes
for STED.
PALM/dSTORM: even with the latest multi-emitter algo-
rithms, the timescale per image depends on the number of
detected molecules, labeling density, etc. Therefore, the trade-
off is, invariably, temporal resolution versus sufficient number
of molecules. There is now considerable effort being put into
particle fluctuation and particle tracking approaches that are
complementary to ‘conventional’ PALM/dSTORM imaging.
FURTHER INFORMATION
Reviews
Blom H, Widengren J. Stimulated emission depletion
microscopy. Chem Rev 2017;117:7377–427. (https://dx.doi.
org/10.1021/acs.chemrev.6b00653)
Cremer C, Masters BR. Resolution enhancement tech-
niques in microscopy. Eur Phys J H 2013;38:281–344. (https://
dx.doi.org/10.1140/epjh/e2012-20060-1)
Eggeling C, Willig KI, Sahl SJ, Hell SW. Lens-based fluo-
rescence nanoscopy. Q Rev Biophys 2015;48:178–243. (https://
dx.doi.org/10.1017/S0033583514000146)
Schermelleh L, Heintzmann R, Leonhardt H. A guide to
super-resolution fluorescence microscopy. JCB 2010;190:165.
(https://dx.doi.org/10.1083/jcb.201002018)
Sheppard C. Fundamentals of superresolution. Micron
2007;38:165–9. (https://dx.doi.org/10.1016/j.micron.2006.07.012)
Sample preparation guide and review for PALM, dSTORM

and related
Allen JR, Ross ST, Davidson MW. Sample preparation for
single molecule localization microscopy. PCCP 2013;15:18771–
83. (https://dx.doi.org/10.1039/c3cp53719f)
Early publications and theoretical background for

the methods
Betzig E, Patterson GH, Sougrat R, et al. Imaging
intracellular fluorescent proteins at nanometer resolu-
tion. Science 2006;313:1642–5. (https://dx.doi.org/10.1126/
science.1127344)
De Luca GM, Breedijk RM, Brandt RA, et al. Re-scan con

focal microscopy: scanning twice for better resolution. BOE
2013;4:2644–56. (https://dx.doi.org/10.1364/BOE.4.002644)
Gustafsson MG. Surpassing the lateral resolution limit by a
factor of two using structured illumination microscopy. J Microsc
2000;198:82–7.(https://dx.doi.org/10.1046/j.1365-2818.2000.00710.x)
Hell SW, Wichmann J. Breaking the diffraction resolution
limit by stimulated emission: stimulated-emission-depletion
fluorescence microscopy. Opt Lett 1994;19:780–2. (https://dx.
doi.org/10.1364/OL.19.000780)
Cox IJ, Sheppard CJR, Wilson T. Super-resolution by con
focal fluorescent microscopy. Optik 1982;60:391–6.
SR-SIM (applications)
Karimi-Ashtiyani R, Ishii T, Niessen M, et al. Point mutation
impairs centromeric CENH3 loading and induces haploid plants.
PNAS 2015;112:11211–6. (https://dx.doi.org/10.1073/pnas.1504333112)
Monteiro JM, Fernandes PB, Vax F, et al. Cell shape dyna-
mics during the staphylococcal cell cycle. Nature Comm 2015;
6:8055. (https://dx.doi.org/10.1038/ncomms9055)
Neumann B, Coakley S, Giordano-Santini R, et al. EFF-1-
mediated regenerative axonal fusion requires components of
the apoptotic pathway. Nature 2015;517:219–22. (https://dx.
doi.org/10.1038/nature14102)
PALM/dSTORM (applications)
Muriel O, Echarri A, Hellriegel C, et al. Phosphorylated
filamin A regulates actin-linked caveolae dynamics. J Cell Sci
2011;124:2763–76. (https://dx.doi.org/10.1242/jcs.080804)
Liebmann T, Blom H, Aperia A, Brismar J. Nanoscale

elucidation of Na,K-ATPase isoforms in dendritic spines. Opt
Nanosc 2013;2:6. (https://dx.doi.org/10.1186/2192-2853-2-6)
Löschberger A, van de Linde S, Debauvalle M-C, et al.
Super-resolution imaging visualizes the eightfold symmetry
of gp210 proteins around the nuclear pore complex and resol-
ves the central channel with nanometer resolution. J Cell Sci
2012;125:570–5. (https://dx.doi.org/10.1242/jcs.098822)
Nan B, Bandaria JM, Moghtaderi A, et al. Flagella stator
homologs function as motors for myxobacterial gliding moti-
lity by moving in helical trajectories. PNAS 2013;110:E1508–13.
(https://dx.doi.org/10.1073/pnas.1219982110)
Airyscan (applications)
Kolossov V, Sivaguru M, Huff J, et al. Airyscan super-
resolution microscopy of mitochondrial morphology and
dynamics in living tumor cells. Mic Res Tech 2016;81:1–14.
(https://dx.doi.org/10.1002/jemt.22732)
Yap CC, Digilio L, McMahon L, Winckler B. The endo
somal neuronal proteins Nsg1/NEEP21 and Nsg2/P19 are
itinerant, not resident proteins of dendritic endosomes. Sci
Rep 2017;7:10481. (https://dx.doi.org/10.1038/s41598-017-
07667-x)
STED (applications)
Bottanelli F, Kromann EB, Allgeyer RS, et al. Two-
colour live-cell nanoscale imaging of intracellular targets.
Nat Commun 2016;7:10778. (https://dx.doi.org/10.1038/
ncomms10778)
Unnersjö-Jess D, Scott L, Blom H, Brismar H. Super-

resolution stimulated emission depletion imaging of slit dia-
phragm proteins in optically cleared kidney tissue. Kidney Int
2015;89:243–7. (https://dx.doi.org/10.1038/ki.2015.308)
Essential
Knowledge
Briefings

Super Resolution Micros

Uploaded by

Copyright:

Available Formats

Super Resolution Micros

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Super Resolution Micros

Uploaded by

Copyright:

Available Formats

Super-resolution

Cover image: microtubules labeled with Alexa-647. The full

About Essential Knowledge Briefings

Super-­resolution techniques can uncover mechanistic details

HISTORY AND BACKGROUND

Figure 1 gives a schematic overview. The techniques, their advan­

Widefield-based approaches Point-scanning approaches

(A) SR-SIM (B) PALM / dSTORM (C) Airyscan (D) STED

Figure 1. Overview schematic of the super-resolution techniques. The image acquisition

are broadly employable and deliver high resolution – albeit not as

while large objects (low frequencies) remain unaffected. Once

Main advantages and limitations of SR-SIM

Localization microscopy (PALM/dSTORM)

optically. The positions can be determined with a precision of

Fluorescence Dark states

mechanism. PALM, developed by Eric Betzig and colleagues

which cause the molecules to switch on and off (hence the

dyes, such as Cy5, that emit outside of the autofluorescence

Advantages and limitations of PALM/dSTORM

dSTORM have limited in vivo applications. Long-­term stabil-

suffering from a high rejection of signal. This is because the

dosage) without signal rejection at closed pinhole. It is a new

Advantages and limitations of Airyscan

and quantifying calcium bursts, photorecovery in diffusion,

Stimulated emission depletion microscopy

limited (see Figure 6 for schematic illustration). The depletion

From the perspective of experimental usability, the chal-

Advantages and limitations of STED

excess of 50 frames per second can be acquired. Larger FOVs

Considerations on sample preparation for all methods

to work (for example, if the dye molecule is not accessible to

CASE STUDY 1. SR-SIM

CASE STUDY 2. STED

Top-view of the filtration slit in podocyte foot-processes from a cleared kidney

CASE STUDY 3. PALM

Overview of a cultured hippocampal neuron genetically expressing the neuronal

CASE STUDY 4. Airyscan

Cells were fixed in 2% paraformaldehyde/3% sucrose/phosphate-buffered saline

PROBLEMS AND SOLUTIONS

medium, at 37°C using, for example, a 63x/1.2 water immer-

to move the grating orientation and frame rate of the camera).

challenging when going into thick, over-labeled, noise-rich,

would perform a raw data quality check as it is being acquired,

Sample preparation guide and review for PALM, dSTORM

Early publications and theoretical background for

De Luca GM, Breedijk RM, Brandt RA, et al. Re-scan con­

Liebmann T, Blom H, Aperia A, Brismar J. Nanoscale

Unnersjö-Jess D, Scott L, Blom H, Brismar H. Super-­

You might also like

Super-resolution techniques can uncover mechanistic details

Figure 1 gives a schematic overview. The techniques, their advan

dSTORM have limited in vivo applications. Long-term stabil-

De Luca GM, Breedijk RM, Brandt RA, et al. Re-scan con

Unnersjö-Jess D, Scott L, Blom H, Brismar H. Super-