Journal of Physics D: Applied Physics
ACCEPTED MANUSCRIPT • OPEN ACCESS
NanoJ: a high-performance open-source super-resolution microscopy
toolbox
To cite this article before publication: Romain F. Laine et al 2019 J. Phys. D: Appl. Phys. in press https://doi.org/10.1088/1361-6463/ab0261
Manuscript version: Accepted Manuscript
Accepted Manuscript is “the version of the article accepted for publication including all changes made as a result of the peer review process,
and which may also include the addition to the article by IOP Publishing of a header, an article ID, a cover sheet and/or an ‘Accepted
Manuscript’ watermark, but excluding any other editing, typesetting or other changes made by IOP Publishing and/or its licensors”
This Accepted Manuscript is © 2019 IOP Publishing Ltd.
 
As the Version of Record of this article is going to be / has been published on a gold open access basis under a CC BY 3.0 licence, this Accepted
Manuscript is available for reuse under a CC BY 3.0 licence immediately.
Everyone is permitted to use all or part of the original content in this article, provided that they adhere to all the terms of the licence
https://creativecommons.org/licences/by/3.0
Although reasonable endeavours have been taken to obtain all necessary permissions from third parties to include their copyrighted content
within this article, their full citation and copyright line may not be present in this Accepted Manuscript version. Before using any content from this
article, please refer to the Version of Record on IOPscience once published for full citation and copyright details, as permissions may be required.
All third party content is fully copyright protected and is not published on a gold open access basis under a CC BY licence, unless that is
specifically stated in the figure caption in the Version of Record.
View the article online for updates and enhancements.
This content was downloaded from IP address 95.85.69.70 on 31/01/2019 at 08:50

----!@#$NewPage!@#$----
  
 
NanoJ: a high-performance open-source 
super-resolution microscopy toolbox 
 
Romain F. Laine1-3 *, Kalina L. Tosheva1,2 *, Nils Gustafsson1,2,4, Robert D. M. Gray1,2,4, Pedro Almada1,2, David Albrecht1, 
Gabriel T. Risa1,6, Fredrik Hurtig7, Ann-Christin Lindås7, Buzz Baum1,6, Jason Mercer1, Christophe Leterrier5, Pedro M. 
Pereira1-3, Siân Culley1-3, and Ricardo Henriques1-3,6 
 
1MRC-Laboratory for Molecular Cell Biology. University College London, London, UK 
2Department of Cell and Developmental Biology, University College London, London, UK 
3The Francis Crick Institute, London, UK 
4Centre for Mathematics and Physics in Life Sciences and Experimental Biology (CoMPLEX), University College London, 
London, UK 
5Aix-Marseille University, CNRS, INP, Institute of Neurophysiopathology, NeuroCyto, Marseille, France 
6Institute for the Physics of Living Systems, University College London, London, UK 
7Department of Molecular Biosciences, The Wenner-Gren Institute, Stockholm University, Stockholm, Sweden 
*These authors contributed equally. 
 
Abstract. Super-resolution microscopy has become essential for the study of nanoscale biological processes. This type of 
imaging often requires the use of specialised image analysis tools to process a large volume of recorded data and extract 
quantitative information. In recent years, our team has built an open-source image analysis framework for super-resolution 
microscopy designed to combine high performance and ease of use. We named it NanoJ - a reference to the popular ImageJ 
software it was developed for. In this paper, we highlight the current capabilities of NanoJ for several essential processing steps: 
spatio-temporal alignment of raw data (NanoJ-Core), super-resolution image reconstruction (NanoJ-SRRF), image quality 
assessment (NanoJSQUIRREL), structural modelling (NanoJ-VirusMapper) and control of the sample environment (NanoJ-
Fluidics). We expect to expand NanoJ in the future through the development of new tools designed to improve quantitative data 
analysis and measure the reliability of fluorescent microscopy studies. 
ImageJ | Fiji | Super-resolution microscopy | Image analysis | Image quality assessment | Fluidics | Resolution | 
Quantitative imaging | Modelling | Virus | Vaccinia | Pox | Archaea | Sulfolobus acidocaldarius 
Correspondence: p.pereira@ucl.ac.uk,s.culley@ucl.ac.uk, r.henriques@ucl.ac.uk 
 
Page 1 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
Introduction. 
Fluorescence 
microscopy 
has 
been 
ubiquitously used in biological studies since its invention 
in the 20th century. It can reveal subcellular structures and 
interactions between specifically labelled molecules and 
allows the quantification of their dynamic behaviour in 
living cells1. Extraction of this biologically relevant 
quantitative information from fluorescence microscopy 
data typically requires digital image processing and 
analysis2. In recent years, Super-Resolution Microscopy 
(SRM) techniques have extended the spatial resolving 
power of fluorescence microscopy beyond the diffraction 
limit3–5. Most SRM techniques use large quantities of raw 
data, often reaching several gigabytes to generate a single 
super-resolution image, thus requiring specialised high-
performance image analysis tools. Several SRM image 
processing 
packages 
are 
available, 
such 
as 
ThunderSTORM6, LAMA7 and SIMcheck8 but each of 
these is focused on a specific type of SRM modality. Here, 
we present NanoJ, a highly versatile set of image 
acquisition and analysis methods developed to improve 
the reliability and quantifiability of microscopy experiments, 
with a particular focus on the demands of live-cell SRM. 
NanoJ is available as a series of ImageJ-based plugins 
which can be used independently or concomitantly. NanoJ 
(Fig. 1) is comprised of the following modules: NanoJ-
Core - general image correction tools including drift 
correction and channel registration, both based on cross-
correlation analysis; NanoJ-SRRF - an analytical 
approach capable of extracting super-resolution data from 
a short sequence of diffraction-limited images, which can 
be 
acquired 
using 
most 
microscopes9,10; 
NanoJ-
SQUIRREL - an algorithm to evaluate resolution and the 
presence of artefacts in super-resolution images11; 
NanoJ-VirusMapper - a single particle analysis method to 
generate nanoscale models of biological structures such 
as viruses12–14; NanoJ-Fluidics - a hardware and software 
framework to control fluidics devices, enabling automation 
of multiplexed experiments15. Thus, the NanoJ framework 
is capable of solving common imaging problems with 
broad biological applications and is compatible with a 
multitude of fluorescence microscope setups and 
experimental protocols. 
 
Fig. 1. NanoJ framework. Currently NanoJ consists of 5 
modules dedicated to super-resolution imaging and 
analysis. 
The NanoJ framework. NanoJ has been designed to 
integrate with the popular ImageJ or Fiji image analysis 
software16,17 and is easily installed as a standard set of 
plugins. NanoJ is also fully open-source and user-friendly. 
The graphical user interfaces (GUIs) are straightforward 
to use and its routines can be easily integrated within 
larger image analysis pipelines through the ImageJ macro 
language. 
NanoJ is designed to be an accessible tool for both 
nonexpert users and developers. Each of its modules 
possess their own separate manual and documentation. 
NanoJ 
is 
implemented 
in 
both 
Java 
(https://www.java.com/) 
and 
OpenCL 
(https://www.khronos.org/opencl), the latter language 
being used for high-performance analysis of image data 
through the use of Graphical Processing Units (GPUs). To 
date, it encompasses four Java ARchive (JAR) packages 
(NanoJ-SRRF, NanoJ-SQUIRREL, NanoJ-VirusMapper, 
NanoJ-Fluidics) that all depend on a central package 
(NanoJ-Core). The core package hosts the libraries that 
enable high-performance GPU-based computing analysis 
and a set of basic image analysis helper methods. The 
modular nature of NanoJ means that its components can 
be updated independently, and the framework can be 
easily extended by appending new analytic packages. 
Page 2 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
NanoJ-Core: Drift Correction. Sample drift commonly 
occurs during the acquisition of SRM data, often as a 
result of gradual changes in the temperature of 
microscope components. Drift introduces motion blur 
artefacts and thereby a loss of resolution. While most 
modern microscopes have an active focus-lock device that 
stabilises the motion of the sample in the axial direction 
(minimising focal drift), the sample will still be prone to 
lateral movement (Fig. 2a). However, in the case where 
the raw data is made up of a sequence of consecutive 
frames acquired rapidly, as is common in SRM methods 
such as Single Molecule Localization Microscopy 
(SMLM)3,4 or fluctuation-based approaches9,18,19, this 
lateral drift can be estimated (Fig. 2b) and analytically 
corrected for each time frame via post-processing (Fig. 2c-
d). NanoJ breaks the task of drift correction into two 
distinct parts: estimation, followed by translation. As a first 
step, NanoJ-Core estimates the linear drift between two 
images by calculating their cross-correlation matrix (CCM) 
(Fig. 2b). The location of the peak intensity in the CCM 
determines the linear shift between the two images, and 
precise sub-pixel accuracy is achieved by up-scaling the 
CCM using a bicubic spline interpolation. Depending on 
the type of acquisition, the reference frame can either be 
the first frame of the raw data or the immediately preceding 
frame. Fig. 2d shows the drift in a 100-frame dataset as 
measured with respect to the first frame. Once drift is 
estimated, the dataset can be directly corrected by 
analytically translating each individual frame using a 
bicubic spline interpolation (Fig. 2c). We estimated the 
performance of NanoJ-Drift correction (see Supp. Note 3 
and Fig. S1a) at varying levels of signal-to-noise ratio 
(SNR) and established that the errors remain within 30 nm 
even at very low SNR and drop quickly below 10 nm for 
SNR > 6, appropriate for super-resolution microscopy. 
The interpolation process will, however, change the noise 
properties of the resulting dataset20, which can have an 
influence 
on 
further 
analysis 
requiring 
specific 
assumptions in noise properties such as Maximum 
Likelihood Estimation (MLE) single-molecule localization 
analyses. 
 
Fig. 2. Drift correction with NanoJ-Core. a) Composite 
image of two frames from a time-lapse dataset of the 
same field-of-view. An artificially large drift was applied 
computationally in order to make it visible for figure 
rendering. b) Cross-correlation map (CCM) between the 
two frames shown in a). The vector position of the 
maximum indicates the linear shift between the two 
frames. c) Overlay of the two frames after drift correction 
using NanoJ-Core. d) Vertical and horizontal drift curves 
obtained using NanoJ-Core from the 100-frame raw data. 
The two images shown in a) correspond to the frames 0 
(green) and 97 (magenta) of the raw data.  
For the specific case of SMLM datasets with sparse 
blinking, there will only be a weak correlation across 
frames, as there is little observable structure conserved 
between consecutive time points. One common strategy 
to alleviate this low correlation is to add fiduciary 
landmarks to the sample, such as static fluorescent beads. 
As an option NanoJ-Core can temporally bin images within 
the dataset, thus increasing the correlation between 
frames and allowing their shift to be more accurately 
estimated21. 
Drift estimation in NanoJ differs from strategies applied by 
other SRM algorithms, such as ThunderSTORM6, by 
analysing unprocessed raw data instead of post-
processed super-resolution reconstructions. This allows 
the estimation to be decoupled from the super-resolution 
image reconstruction algorithm and hence drift-corrected 
raw data can be analysed using a wider range of methods 
including SRRF9 and SOFI18. Furthermore, NanoJ-SRRF 
(as described in a following section) can import the drift 
curve (Fig. 2d) created by NanoJ-Core Drift Correction 
Page 3 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
and use this information directly during analysis without 
the need to pre-translate each frame in the raw dataset. 
NanoJ-Core: Channel registration. In multicolour 
fluorescence microscopy, images acquired in different 
spectral channels are often misaligned as a result of 
chromatic aberrations in the optical path and the use of 
different filter sets for each colour. This misalignment is 
frequently ignored in conventional microscopy as it 
typically occurs on a scale smaller than the diffraction limit. 
However, this effect becomes nonnegligible in the context 
of SRM22. Channel registration is therefore essential for 
multicolour SRM studies quantifying colocalization or 
interactions between different structures23–25. Advanced 
imaging research groups tend to develop their own 
analysis scripts for their particular applications, and thus 
no user-friendly tools are readily available. Therefore, 
NanoJ-Core Channel registration offers a unique tool to 
perform this registration easily and robustly on a wide 
variety of datasets. 
The shift between different spectral channels is usually 
inhomogeneous across a field of view, which prevents the 
use of typical CCM-based approaches such as the one 
described for NanoJ-Core Drift Correction. In order to 
characterise the spectral misalignment across a field of 
view, it is necessary to image a sample where the same 
structure can be observed across all spectral channels of 
interest. Furthermore, the sample should contain 
structures occupying the whole field of view. A typical 
sample for this characterisation is a coverslip coated with 
a large number of beads labelled with multiple fluorescent 
dyes26. Following multicolour imaging of this sample, 
NanoJ-Core can be used to calculate a nonlinear two-
dimensional spatial transform describing the misalignment 
for each channel relative to a reference channel. Given 
that chromatic misalignment is a fixed property of an 
optical system, NanoJ-Core can then apply these 
transforms to realign other multicolour datasets acquired 
using the same optical path (Fig. 3)27,28. 
To generate these non-linear misalignment fields NanoJ-
Core first calculates local, linear misalignments between 
channels (Fig. 3a). This is achieved by dividing the image 
into small areas ("blocks"). For each block, the shift is 
calculated by finding the cross-correlation peak position as 
shown in Fig. 2a, b. These local shift values are then 
interpolated across the whole field-of-view using an 
inverse distance weighting interpolation29. This generates 
two smooth shift maps describing the misalignments in the 
horizontal and vertical directions (Fig. 3b). For a given 
channel, each pixel value within the horizontal/vertical shift 
map indicates the horizontal/vertical displacement that 
needs to be applied to that pixel to align it with the 
reference channel. NanoJ-Core can then be used to apply 
these maps to align channels in the dataset of interest (Fig. 
3), provided they have been acquired using the same 
optical configuration. Here, we measured the Target 
Registration Error28,30 (TRE, see Supp. Note 3 and Fig. 
S1b) in typical multi-colour configurations and showed that 
the TRE remains below 50 nm for all combinations of 
channels. 
NanoJ-Core performs the channel registration by creating 
a new image representing each channel, where the 
intensity value for each pixel coordinate corresponds to 
the intensity value from the original image at the 
equivalent coordinate corrected for local shift. For cases 
in which these coordinates are not discrete (sub-pixel 
shift), a bicubic spline interpolation is used to recover pixel 
values in continuous space. Because the shift map can be 
extrapolated to continuous space, the registration 
procedure obtained from diffraction-limited images can 
also easily be used to correct super-resolved images 
obtained using the same optical configuration. 
NanoJ-SRRF: Live-Cell Super-Resolution Imaging. As 
part of the NanoJ framework, we include our recently 
developed 
SRM 
reconstruction 
algorithm 
Super-
Resolution Radial Fluctuations (SRRF), which is able to 
extract sub-diffraction information from a short burst of 
images acquired at highspeed with modern fluorescence 
microscopes9,10. SRRF is a purely analytical approach. It 
alleviates the need to use toxic photoswitching-inducing 
buffers31, specialised fluorophores32,33, damaging high-
Page 4 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
intensity illumination34 or specialised equipment5,35 when 
compared to other SRM methods 3–5,35. 
 
Fig. 3. Multi-colour channel registration with NanoJ-
Core. a) Composite image of multi-colour TetraSpeck™ 
beads imaged in two different channels ("GFP-channel" 
indicated in green and "mCherry-channel" in magenta), 
prior to (left) and after (right) channel registration using 
NanoJ-Core. Insets - individual beads from indicated 
locations. Scale bars: 25 µm, insets: 0.5 µm. b) Vectorial 
representation of the shift between the two channels (left, 
displacement 
vector 
length 
50 
times 
larger 
for 
representation purposes), horizontal (middle) and vertical 
(right) shift maps obtained and applied to the data shown 
in a). Scale bars: 25 µm. 
SRRF is based on similar principles to SMLM, with the key 
difference that it does not rely on the detection of 
spatiotemporally isolated fluorophores. Instead, SRRF 
generates a magnified pixel grid where each pixel value 
relates to the probability of fluorophores existing in that 
corresponding region of space. To do this, SRRF 
calculates the local radial symmetry (’radiality’) in each 
pixel of the magnified image using local intensity gradient 
information. The obtained radiality value will be high when 
a point-spread-function (PSF) profile transiently becomes 
dominant, highlighting the presence of a fluorescent 
molecule at that location. Furthermore, the fluctuations of 
radiality values follow the underlying natural intensity 
fluctuations of fluorophores, which have a distinct 
temporal signature to that of noise18.  
Therefore, a temporal correlation of the radiality at each 
pixel can be projected into a final image, where the 
structures of interest will be better resolved.  
Thanks to its low-illumination requirement, SRRF is highly 
suited to enable live-cell SRM and allows for long SRM 
timelapse acquisitions10. Fig. 4 shows a typical live-cell 
SRRF acquisition of a Cos7 cell expressing UtrCH-GFP (a 
probe for actin filaments) imaged at 33.3 Hz. The dataset 
shown is part of a longer > 30 min time-course dataset 
(Sup. Movie 1 will be added soon). SRRF allows the 
observation of protein dynamics (e.g. actin) during long 
periods with no perceivable phototoxicity. 
 
Fig. 4. Live-cell super-resolution microscopy with 
NanoJ-SRRF. a) Comparison of widefield (left) and SRRF 
reconstruction (right) obtained from a Cos7 cell 
expressing UtrCH-GFP to label actin filaments. Scale bar: 
5 µm. b) Time-course of the inset shown in a), obtained 
from a continuous imaging at 30 ms exposure (33.3 Hz) 
and displayed every 30 s. Scale bar: 1 µm. c) Colour-
coded time course dataset from b). Scale bar: 1 µm. 
NanoJ-SRRF, the software implementation of the SRRF 
algorithm, uses GPU computing whenever possible to 
accelerate radial symmetry calculations. This is achieved 
by implementing part of its analytic engine in OpenCL, with 
a fallback of execution to the CPU if a compatible graphics 
card is not detected. 
Page 5 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
NanoJ-SQUIRREL: Estimating Image Quality. SRM 
techniques 
are 
more 
complex 
than 
conventional 
diffraction-limited microscopy. This complexity arises from 
the sample preparation requirements (SMLM), the 
microscope hardware (stimulated emission depletion 
microscopy or STED, Structured Illumination Microscopy 
or SIM) and the post-acquisition image processing (SMLM, 
SIM). Unsuitable choices in any of these underlying 
variables can lead to artefacts within the final image and 
the potential for false conclusions to be drawn. NanoJ-
SQUIRREL (Super-resolution Quantitative Image Rating 
and Reporting of Error Locations) is an algorithm that 
highlights the presence of artefacts in super-resolution 
images. It does so by calculating quantitative maps 
showing both local SRM image quality and resolution11 
and can thus be used to optimise acquisition workflows10. 
The central concept of SQUIRREL is that a diffraction-
limited image and the corresponding SRM image of the 
same region should contain the same underlying structure, 
just at different resolutions. Thus, the diffraction-limited 
image can be treated as a high-confidence benchmark 
against which the SRM image can be compared. The 
SQUIRREL 
image 
quality 
assessment 
algorithm 
formalises this comparison analytically. 
NanoJ-SQUIRREL quality assessment requires the user 
to provide an acquired diffraction-limited image and the 
SRM image of the same structure (Fig. 5a). As both 
images represent the underlying fluorophore distribution 
but at different resolutions, there exists a blurring function 
that can convert the SRM image into its diffraction-limited 
equivalent. SQUIRREL calculates this blurring function 
and applies it to the SRM image. The blurred image is then 
compared against the diffraction-limited reference. This 
generates three quality indicators. An error map is 
produced that maps the discrepancy between the blurred 
SRM image and the reference image at every pixel (Fig. 
5b). This highlights regions where the super-resolution 
image is inconsistent with the reference image. For 
example, the three insets in Fig. 5b show typical SMLM 
artefacts 
including 
incomplete 
structures 
and 
interstructure mislocalization. Two global quality metrics 
are also generated for each image: the RSP (Resolution 
Scaled Pearson’s Correlation Coefficient) and the RSE 
(Resolution Scaled Error). The RSP can take a value in 
the interval [-1,1] and describes the structural agreement 
between the blurred super-resolution and reference 
images. Here, higher values indicate better agreement, 
with an RSP of 1 indicating a perfect structural match. The 
RSE describes the mean intensity mismatch between the 
blurred super-resolution and reference images; in this 
case lower values represent better agreement with a value 
of 0 indicating a perfect intensity match. 
NanoJ-SQUIRREL: Estimating Image Resolution. The 
purpose of SRM is to resolve finer structural detail than is 
achievable 
with 
conventional 
diffraction-limited 
microscopy. It is therefore useful to have an objective 
measurement of resolution within a super-resolution 
image, for example to enable meaningful analysis and 
structural hypotheses that stay in line with the actual 
precision of the data. The current standard for measuring 
image resolution in SMLM images is Fourier Ring 
Correlation (FRC)36. This method involves comparing two 
independently acquired super-resolution images of the 
same field-of-view so that they only differ by their noise 
components. For SMLM datasets the two SRM images are 
usually obtained by splitting localizations from odd and 
even frames. The correlation between these two SRM 
images is measured at different frequencies in Fourier 
space; the frequency at which this correlation drops below 
a set threshold indicates the resolution of the image. FRC 
has been previously implemented in ImageJ36, but only 
gives a single resolution measurement for the entire field 
of 
view. 
However, 
resolution 
is 
not 
necessarily 
homogeneous across the SRM image (Fig. 5c). This is 
particularly true for SMLM methods as localization 
accuracy depends strongly on labelling density and laser 
illumination intensity, which can both vary considerably 
within a single field of view.  
Page 6 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
 
Fig. 5. Quality assessment and resolution mapping 
with NanoJ-SQUIRREL. a) A super-resolution rendering 
(left) and acquired widefield image (right) of fixed 
microtubules labelled with Alexa Fluor-647. b) Left: 
SQUIRREL error map highlighting discrepancies between 
the super-resolution and diffraction-limited images in (a). 
Right: Magnified insets of super-resolution rendering at 
indicated positions on error map. c) Left: SQUIRREL 
resolution map of the super-resolution image in (a). Right: 
Magnified insets of super-resolution rendering for 
indicated resolution blocks. Whole image scale bars = 
5µm, inset scale bars = 1µm. 
Furthermore, FRC can generate biased measurements for 
certain fluorophore distributions such as point-like patterns. 
Therefore, an additional feature of the NanoJ-SQUIRREL 
plugin is local mapping of FRC resolution across an image. 
To do this, the user provides an image stack comprising 
two independent renderings of the same dataset (e.g. 
through the odd/even frames splitting as described above). 
The images are then spatially split into equal-sized blocks 
and FRC analysis is run locally on each block. For blocks 
where there is insufficient correlation to generate an FRC 
resolution value, a resolution value is interpolated from 
neighbouring blocks. Fig. 5c shows the FRC map obtained 
from the SRM image shown in (a). This map highlights that 
the resolution in this image varies between 85 and 36 nm. 
It is important to note that high resolution (that is, a low 
FRC value) does not imply that the super-resolution image 
has depicted structures correctly; it only means that there 
is low variation in the locations of the fluorophores 
between the two rendered images. Therefore, the error 
mapping functionality within NanoJ-SQUIRREL can 
complement FRC-mapping in order to obtain a more 
complete perspective on super-resolution image quality. 
NanoJ-VirusMapper: 
Structural 
Mapping 
and 
Modelling. As part of the NanoJ framework, we include a 
unique single-particle analysis (SPA) tool called NanoJ-
VirusMapper. It is the first open-source, freely available 
algorithm 
for 
unbiased, 
high-throughput 
SPA 
of 
fluorescence imaging and allows the structural modelling 
of viruses and other macromolecular complexes 12–14. The 
principle of SPA is to image many identical copies of a 
structure, independently of its orientation, and align and 
combine them to build an averaged structural map of the 
underlying structure with high signal-to-noise ratio37–41. 
The SPA implementation of VirusMapper facilitates 
automatic processing of multiple images to detect, 
segment, align, classify and average thousands of 
individual structures. It is entirely general, assuming no 
underlying symmetry or other properties of the imaged 
structure. Here, we illustrate this with models of three 
distinct vaccinia virus substructures42: core, lateral bodies 
(LBs) and membrane. All three substructures were 
labelled on viral particles and imaged with SIM (Fig. 6a, 
top). VirusMapper can explicitly incorporate information 
from multiple fluorescence channels to enable multi-
component modelling. Here, the images of the LBs and 
the core were used to find the orientations of each particle 
in the dataset. This allowed us to create simultaneous 
models of the three components of the virus (Fig. 6a, 
Page 7 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
bottom). VirusMapper can also be used to separately 
model different 3D orientations of a structure.  
 
Fig. 6. Quantitative SPA-based modelling with NanoJ-
VirusMapper. a) Top: Aligned SIM images of individual 
vaccinia particles labelled for L4 (core), F17 (lateral bodies, 
LBs) and A17 (membrane, mem.). Bottom: VirusMapper 
models of the three channels. Scale bars: 200nm. b) Top: 
Aligned 
SIM 
images 
of 
individual 
Sulfolobus 
acidocaldarius cells labelled for the S-layer (S) and the 
archaeal ESCRT-III homolog CdvB (CdvB). Bottom: 
VirusMapper models of three different orientations of the 
cells; magenta - S-layer, green - CdvB. Scale bars: 1µm. 
Here, we highlight this capability on a study of the division 
ring in the thermoacidophilic archaeon Sulfolobus 
acidocaldarius, strain DSM639 (Fig. 6b). Sulfolobus cells 
were labelled for both their outer S-layer and the division 
ring, as marked by the archaeal ESCRT-III homolog CdvB. 
The images of the division ring were used to identify 
different orientations of the cells (Fig. 6b) and separate 
templates were created using VirusMapper. Parallel 
models were then generated of the division ring and the 
Slayer using these templates for each orientation. The 
models obtained from the two channels can then be 
overlayed (Fig. 6b, bottom). VirusMapper is compatible 
with any fluorescence microscopy method and has been 
demonstrated with SIM, STED microscopy12 and SMLM 
(Fig. S1). The general approach of the method enables it 
to even be applied to correlative combinations of methods 
(Fig. S1). 
NanoJ-Fluidics: Sample Liquid Exchange. NanoJ-
Fluidics is a hardware and software framework for precise 
and accurate automated liquids exchange15. It was 
developed to enable automation of sample treatment and 
labelling of live or fixed specimens directly on the 
microscope stage15,43. The NanoJ-Fluidics hardware 
component is composed of customisable, low-cost and 
robust LEGO® syringe pumps and a liquid removal 
peristaltic pump, all controlled by simple Arduino® 
electronics. It is compatible with off the-shelf imaging 
chambers, without the need for any microfabrication. Its 
control software (the NanoJ-Fluidics module) is ImageJ-
based and can be fully integrated with microscopy 
acquisition 
software. We 
have 
demonstrated 
the 
applicability of NanoJ-Fluidics in multiple experimental 
contexts, including in-situ correlative live-to-fixed super-
resolution imaging, multimodal super-resolution imaging 
and event-driven fixation15. The approach can also be 
easily extended to protocol optimisation (e.g. titrating 
antibody concentrations or adjusting imaging buffer 
composition) or liquid exchange protocols integrated with 
the imaging (e.g. drug delivery or automated event-driven 
fixation). 
Here, we demonstrate NanoJ-Fluidics by acquiring a high 
quality multicolour SMLM dataset, where STORM and 
DNA-PAINT44 imaging strategies are combined into a 
single workflow (Fig. 7a). This approach is particularly 
suited to multi-target imaging 32, which is difficult to 
achieve with standard sample preparation techniques due 
to the low number of suitable fluorophores available for 
SMLM 
microscopy. With 
NanoJ-Fluidics, 
we 
can 
seamlessly perform all labelling steps in an automated and 
reliable manner directly on the microscope stage. We 
Page 8 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
showcase this using a four-channel acquisition of actin 
with STORM, and mitochondria, vimentin and clathrin with 
DNA-PAINT 
(Fig. 
7b). 
NanoJ-Fluidics’ 
highly 
customisable nature has already spawned several 
alternative 
designs 
from 
the 
community 
(https://github.com/HenriquesLab/NanoJ-Fluidics/wiki), 
which are in constant development. NanoJ-Fluidics makes 
imaging 
protocol 
automation 
readily 
available 
to 
researchers, hence improving not only the reliability and 
repeatability of the protocols, but also the range of 
protocols that are achievable. 
 
Fig. 7. Automated DNA-PAINT and STORM imaging. a) 
NanoJ-Fluidics workflow used for multi-color STORM and 
DNA-PAINT imaging. b) Left: 4-channel merge of STORM 
and DNA-PAINT with actin (red, STORM), mitochondria 
(green, DNAPAINT I1 strand), vimentin (yellow, DNA-
PAINT I2 strand) and clathrin (cyan, DNAPAINT I3 strand). 
Right: Single-channel images from the inset. Scale bars: 
2 µm. 
Discussion and Future Perspectives. The NanoJ 
framework provides a unique and comprehensive set of 
tools to support fluorescence imaging, from data 
acquisition and protocol optimisation to structural 
quantification. It offers powerful and user-friendly solutions 
to common pitfalls of image analysis such as drift 
correction and channel registration (NanoJ-Core). It 
extends the tools available for SRM to expert and novice 
users alike and aligns with the general aim of the 
community to extend SRM to live-cell applications, offering 
the first open-source analytical approach for live cell SRM 
imaging (NanoJ-SRRF). NanoJ offers a way to correlate 
dynamic live-cell data with high-resolution fixed cell data 
in an automated and highly reproducible manner (NanoJ-
Fluidics). It incorporates an analytical method for the 
generation of accurate, high-content, molecular specific 
models 
(NanoJ-VirusMapper). 
Finally, 
the 
NanoJ 
framework provides tools to quantitatively assess SRM 
image quality by spatially mapping unbiased quality 
metrics, artefacts (errors) and FRC resolution (NanoJ-
SQUIRREL), with the aim to improve standards in 
assessing and reporting microscopy data. 
The NanoJ framework gives access user-friendly yet 
robust analytical methods in tandem with new approaches 
to perform live-cell SRM experiments. A typical SRM live-
cell imaging experiment using the NanoJ framework would 
entail acquiring a SRRF-compatible raw data set in any 
modern 
fluorescence 
microscope; 
performing 
drift 
correction and channel registration; followed by SRRF 
analysis for SRM image reconstruction; and finally, quality 
check using SQUIRREL. Used concomitantly in this 
manner, these tools can be used to address a large variety 
of biological questions and can be seamlessly repeated 
for imaging protocol optimisation and complemented using 
NanoJ-VirusMapper and/or NanoJ-Fluidics. Thus, the 
NanoJ-enabled scientific pipeline allows any user, 
regardless of the experience level, to seamlessly obtain 
SRM quantitative data of the highest scientific quality. The 
NanoJ framework was developed to facilitate access and 
expand the options of research groups using fluorescence 
microscopy. Hence, speed, reliability, performance and 
cross compatibility are of paramount importance. In this 
context, NanoJ has been designed for high-performance 
image analysis, using GPU computing, ensuring the quick 
processing of large data volumes45–49. Further, NanoJ’s 
modular and open-source nature and its integration within 
ImageJ allow it to be used in conjunction with other 
analysis software packages50–53 (49–52). We are 
continuously supporting, adapting and expanding the 
framework to include new approaches, such as 3D 
imaging. 
Page 9 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
We expect NanoJ to set the standard for useful, open-
source, high performance methods for the whole 
microscopy community. 
Software and Hardware Availability. NanoJ follows 
opensource software and hardware standards. Each of its 
modules can be installed by enabling the corresponding 
code repository in Fiji or by following the instructions on 
the corresponding websites: 
• https://github.com/HenriquesLab/NanoJ-Core 
• https://github.com/HenriquesLab/NanoJ-SRRF 
• https://bitbucket.org/rhenriqueslab/NanoJ-
SQUIRREL 
• https://bitbucket.org/rhenriqueslab/NanoJ-
VirusMapper 
• https://github.com/HenriquesLab/NanoJ-Fluidics 
 
ACKNOWLEDGEMENTS 
First and foremost, we would like to thank the ImageJ and 
Fiji 
open-source 
development 
community. 
Their 
astonishing work has been the inspiration to the creation 
of NanoJ. There has been a large number of beta-testers 
and NanoJ users whose feedback helped us to continually 
update and improve its code, to the best of our capacity 
we keep an updated list of significant contributors in the 
website corresponding to each NanoJ module. We thank 
Prof. Ralf Jungmann at Max Planck Institute of 
Biochemistry Munich for reagents and advice. This work 
was funded by grants from the UK Biotechnology and 
Biological Sciences Research Council (BB/M022374/1; 
BB/P027431/1; BB/R000697/1; BB/S507532/1) (R.H., 
P.M.P. and R.F.L.), the UK Medical Research Council 
(MR/K015826/1) 
(R.H.), 
the 
Wellcome 
Trust 
(203276/Z/16/Z) (S.C., R.H and B.B.), Core funding to the 
MRC Laboratory for Molecular Cell Biology, University 
College London (MC_UU12018/7) (J.M.), the European 
Research Council (649101—UbiProPox) (J.M.) and the 
Centre National de la Recherche Scientifique (CNRS 
ATIP-AVENIR program AO2016) (C.L.). N.G. and 
R.D.M.G funded by the Engineering and Physical 
Sciences Research Council (EP/L504889/1). P.A. was 
supported by a PhD fellowship from the UK’s 
Biotechnology and Biological Sciences Research Council. 
Research by B.B. was supported by UCL, Cancer 
Research UK (C1529/A17343), BBSRC (BB/P001440/1) 
and MRC (MC_CF12266). K.L.T. and G.T.R. are 
supported by a 4-year MRC Research Studentship. D.A. 
is presently a Marie Curie fellow (Marie Sklodowska-Curie 
750673). A.C.L. AND F.H. were supported by the Swedish 
Research Council, grant 621-20134685. 
 
AUTHOR CONTRIBUTIONS 
These 
contributions 
follow 
the 
Contributor 
Roles 
Taxonomy 
guidelines: 
https://casrai.org/credit/. 
Conceptualization: K.L.T, R.F.L., N.G., R.D.M.G, P.A., 
D.A., G.T.R., B.B., J.M., C.L., P.M.P., S.C., R.H.; Data 
curation: K.L.T, R.F.L., R.D.M.G, G.T.R., J.M., C.L., 
P.M.P., S.C., R.H.; Formal analysis: K.L.T, R.F.L., 
R.D.M.G, C.L., S.C.; Funding acquisition: A.C.L, B.B., 
C.L., R.H.; Investigation: K.L.T, R.F.L., R.D.M.G, G.T.R., 
C.L., P.M.P., S.C.; Methodology: R.F.L., N.G., R.D.M.G, 
P.A., J.M., C.L., P.M.P., S.C., R.H.; Project administration: 
R.F.L., B.B., J.M., C.L., P.M.P., S.C., R.H.; Resources: 
K.L.T, R.D.M.G, D.A., G.T.R., F.H., A.C.L., B.B., J.M., C.L., 
P.M.P., S.C.; Software: R.F.L., N.G., R.D.M.G, P.A., C.L., 
P.M.P., S.C., R.H.; Supervision: R.F.L., A.L., B.B., J.M., 
C.L., P.M.P., S.C., R.H.; Validation: 
K.L.T, R.F.L., N.G., R.D.M.G, P.A., D.A., C.L., P.M.P., 
S.C.; Visualization: K.L.T, R.F.L., R.D.M.G, D.A., C.L., 
P.M.P., S.C., R.H.; Writing – original draft: K.L.T, R.F.L., 
R.D.M.G, D.A., C.L., P.M.P., S.C., R.H.; Writing – review 
& editing: all authors. 
COMPETING FINANCIAL INTERESTS The authors 
declare no competing financial interests. 
Page 10 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
Bibliography 
1. 
Rino, J., Braga, J., Henriques, R. & Carmo-Fonseca, M. Frontiers in fluorescence microscopy. Int. J. Dev. Biol. 53, 
1569–1579 (2009). 
2. 
Wheeler, A. & Henriques, R. Standard and Super-resolution Bioimaging Data Analysis: A Primer. (John Wiley & Sons., 
2017). 
3. 
Betzig, E. et al. Imaging intracellular fluorescent proteins at nanometer resolution (SOM). Science (80-. ). x, 1642–1646 
(2006). 
4. 
Rust, M. J., Bates, M. & Zhuang, X. Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy 
(STORM). Nat. Methods 3, 793–796 (2006). 
5. 
Hell, S. W. & Wichmann, J. Breaking the diffraction resolution limit by stimulated emission: stimulated-emission-depletion 
fluorescence microscopy. Opt. Lett. 19, 780 (1994). 
6. 
Ovesný, M. et al. ThunderSTORM: A comprehensive ImageJ plug-in for PALM and STORM data analysis and super-
resolution imaging. Bioinformatics 30, 2389–2390 (2014). 
7. 
Malkusch, S. & Heilemann, M. Extracting quantitative information from single-molecule super-resolution imaging data 
with LAMA – LocAlization Microscopy Analyzer. Sci. Rep. 6, (2016). 
8. 
Schermelleh, L. et al. SIMcheck: a toolbox for successful super-resolution SIM imaging. Sci. Rep. 1–11 (2015). 
9. 
Gustafsson, N. et al. Fast live-cell conventional fluorophore nanoscopy with ImageJ through super-resolution radial 
fluctuations. Nat. Commun. 7, 12471 (2016). 
10. 
Culley, S., Tosheva, K. L., Matos Pereira, P. & Henriques, R. SRRF: Universal live-cell super-resolution microscopy. Int. 
J. Biochem. Cell Biol. 101, 74–79 (2018). 
11. 
Culley, S. et al. Quantitative mapping and minimization of super-resolution optical imaging artifacts. Nat. Methods (2018). 
doi:10.1038/nmeth.4605 
12. 
Gray, R. D. M. et al. VirusMapper: open-source nanoscale mapping of viral architecture through super-resolution 
microscopy. Sci. Rep. 6, 29132 (2016). 
13. 
Gray, R. D. M., Mercer, J. & Henriques, R. Open-source Single-particle Analysis for Super-resolution Microscopy with 
VirusMapper. J. Vis. Exp. 1–10 (2017). doi:10.3791/55471 
14. 
Gray, R. et al. Nanoscale Polarization of the Vaccinia Virus Entry Fusion Complex Drives Efficient Fusion. bioRxiv 
360073 (2018). doi:10.1101/360073 
15. 
Almada, P. et al. Automating multimodal microscopy with NanoJ-Fluidics. bioRxiv 320416 (2018). doi:10.1101/320416 
16. 
Abramoff, M. D., Magalhães, P. J. & Ram, S. J. Image Processing with ImageJ. Biophotonics Int. 11, 36–42 (2004). 
17. 
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012). 
18. 
Dertinger, T., Colyer, R., Iyer, G., Weiss, S. & Enderlein, J. Fast, background-free, 3D super-resolution optical fluctuation 
imaging (SOFI). Proc. Natl. Acad. Sci. 106, 22287–22292 (2009). 
19. 
Cox, S. et al. Bayesian localization microscopy reveals nanoscale podosome dynamics. Nat. Methods 9, 195–200 (2011). 
20. 
Blaysat, B., Grédiac, M. & Sur, F. Effect of interpolation on noise propagation from images to DIC displacement maps. 
Int. J. Numer. Methods Eng. 108, 213–232 (2016). 
21. 
Mlodzianoski, M. J. et al. Sample drift correction in 3D fluorescence photoactivation localization microscopy. Opt. 
Express 19, 15009 (2011). 
22. 
Erdelyi, M. et al. Correcting chromatic offset in multicolor super-resolution localization microscopy. Opt. Express 21, 
10978 (2013). 
23. 
Bock, H. et al. Two-color far-field fluorescence nanoscopy based on photoswitchable emitters. Appl. Phys. B 88, 161–
165 (2007). 
24. 
Van De Linde, S. et al. Multicolor photoswitching microscopy for subdiffraction-resolution fluorescence imaging. 
Photochem. Photobiol. Sci. 8, 465–469 (2009). 
25. 
Niekamp, S., Sung, J., Huynh, W., Vale, R. D. & Stuurman, N. High accuracy measurements of nanometer-scale 
distances between fluorophores at the single-molecule level. bioRxiv 234740 (2017). doi:10.1101/234740 
26. 
Demmerle, J. et al. Strategic and practical guidelines for successful structured illumination microscopy. Nat. Protoc. 12, 
Page 11 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
988–1010 (2017). 
27. 
Arganda-Carreras, I. et al. Consistent and Elastic Registration of Histological Sections Using Vector-Spline 
Regularization BT  - Computer Vision Approaches to Medical Image Analysis. in (eds. Beichel, R. R. & Sonka, M.) 85–
95 (Springer Berlin Heidelberg, 2006). 
28. 
Annibale, P., Scarselli, M., Greco, M. & Radenovic, A. Identification of the factors affecting co-localization precision for 
quantitative multicolor localization microscopy. Opt. Nanoscopy 1, 9 (2012). 
29. 
Shepard, D. A two-dimensional interpolation for irregularly-spaced data function. Proc. Ass. Comput. Mach. 517–524 
(1968). doi:10.1145/800186.810616 
30. 
Churchman, L. S., Okten, Z., Rock, R. S., Dawson, J. F. & Spudich, J. A. Single molecule high-resolution colocalization 
of Cy3 and Cy5 attached to macromolecules measures intramolecular distances through time. Proc. Natl. Acad. Sci. 
102, 1419–1423 (2005). 
31. 
Henriques, R., Griffiths, C., Rego, E. H. & Mhlanga, M. M. PALM and STORM: Unlocking live-cell super-resolution. 
Biopolymers 95, 322–331 (2011). 
32. 
Dempsey, G. T., Vaughan, J. C., Chen, K. H., Bates, M. & Zhuang, X. Evaluation of fluorophores for optimal performance 
in localization-based super-resolution imaging. Nat. Methods 8, 1027–1036 (2011). 
33. 
Henriques, R. & Mhlanga, M. M. PALM and STORM: What hides beyond the Rayleigh limit? Biotechnol. J. 4, 846–857 
(2009). 
34. 
Wäldchen, S., Lehmann, J., Klein, T., van de Linde, S. & Sauer, M. Light-induced cell damage in live-cell super-resolution 
microscopy. Sci. Rep. 5, 15348 (2015). 
35. 
Gustafsson, M. G. L. Surpassing the lateral resolution limit by a factor of two using structured illumination microscopy. 
SHORT COMMUNICATION. J. Microsc. 198, 82–87 (2000). 
36. 
Nieuwenhuizen, R. P. J. et al. Measuring image resolution in optical nanoscopy. Nat. Methods 10, 557–62 (2013). 
37. 
Szymborska, A. et al. Nuclear pore scaffold structure analyzed by super-resolution microscopy and particle averaging. 
Science 341, 655–8 (2013). 
38. 
Laine, R. F. et al. Structural analysis of herpes simplex virus by optical super-resolution imaging. Nat. Commun. 6, 5980 
(2015). 
39. 
Lelek, M. et al. Superresolution imaging of HIV in infected cells with FlAsH-PALM. Proc. Natl. Acad. Sci. 109, 8564–
8569 (2012). 
40. 
Verdier, T., Gunzenhauser, J., Manley, S. & Castelnovo, M. Single particle maximum likelihood reconstruction from 
superresolution microscopy images. PLoS One 12, e0172943 (2017). 
41. 
Sieben, C., Banterle, N., Douglass, K. M., Gönczy, P. & Manley, S. Multicolor single-particle reconstruction of protein 
complexes. Nat. Methods 15, 777–780 (2018). 
42. 
Moss, B. in Fields Virology 6 ed. (eds. Knipe, D. M. & Howley, P.) 2664 (Philadelphia, USA: Lippincott Williams & Wilkins, 
a Wolters Kluwer business, 2013). 
43. 
Dix, C. L. et al. The Role of Mitotic Cell-Substrate Adhesion Re-modeling in Animal Cell Division. Dev. Cell 45, 132–
145.e3 (2018). 
44. 
Jungmann, R. et al. Multiplexed 3D cellular super-resolution imaging with DNA-PAINT and Exchange-PAINT. Nat. 
Methods 11, 313–8 (2014). 
45. 
Herbert, S., Soares, H., Zimmer, C. & Henriques, R. Single-Molecule Localization Super-Resolution Microscopy: Deeper 
and Faster. Microsc. Microanal. 18, 1419–1429 (2012). 
46. 
Pereira, P. M., Almada, P. & Henriques, R. High-content 3D multicolor super-resolution localization microscopy. Methods 
Cell Biol. 125, 95–117 (2015). 
47. 
Almada, P., Culley, S. & Henriques, R. PALM and STORM: Into large fields and high-throughput microscopy with 
sCMOS detectors. Methods 88, 109–121 (2015). 
48. 
Beghin, A. et al. Localization-based super-resolution imaging meets high-content screening. Nat. Methods 14, 1184–
1190 (2017). 
49. 
Douglass, K. M., Sieben, C., Archetti, A., Lambert, A. & Manley, S. Super-resolution imaging of multiple cells by 
optimized flat-field epi-illumination. Nat. Photonics 10, 705–708 (2016). 
Page 12 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript

----!@#$NewPage!@#$----
  
 
50. 
Sage, D. et al. Super-resolution fight club: A broad assessment of 2D & 3D single-molecule localization microscopy 
software 2. (2018). doi:10.1101/362517 
51. 
Weigert, M. et al. Content-Aware Image Restoration: Pushing the Limits of Fluorescence Microscopy. 1–16 (2017). 
doi:10.1101/236463 
52. 
Henriques, R. et al. QuickPALM: 3D real-time photoactivation nanoscopy image processing in ImageJ. Nat. Methods 7, 
339–340 (2010). 
53. 
Laine, R. F. et al. MiLeSIM: combining super-resolution and machine learning permits high-throughput virus structure 
analysis. bioRxiv (2018). 
 
Page 13 of 13
AUTHOR SUBMITTED MANUSCRIPT - JPhysD-118607.R1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Accepted Manuscript