Cover
Title Page
List of Contributors
Foreword
1 Digital Microscopy
1. 1.1 ACQUISITION
2. 1.2 INITIALISATION
3. 1.3 MEASUREMENT
4. 1.4 INTERPRETATION
5. 1.5 REFERENCES
2 Quantification of Image Data
1. 2.1 MAKING SENSE OF IMAGES
2. 2.2 QUANTIFIABLE INFORMATION
3. 2.3 WRAPPING UP
4. 2.4 REFERENCES
3 Segmentation in Bioimaging
1. 3.1 SEGMENTATION AND INFORMATION CONDENSATION
2. 3.2 EXTRACTING OBJECTS
3. 3.3 WRAPPING UP
4. 3.4 REFERENCES
4 Measuring Molecular Dynamics and Interactions by Förster Resonance Energy Transfer (FRET)
1. 4.1 FRET‐BASED TECHNIQUES
2. 4.2 EXPERIMENTAL DESIGN
3. 4.3 FRET DATA ANALYSIS
4. 4.4 COMPUTATIONAL ASPECTS OF DATA PROCESSING
5. 4.5 CONCLUDING REMARKS
6. 4.6 REFERENCES
5 FRAP and Other Photoperturbation Techniques
1. 5.1 PHOTOPERTURBATION TECHNIQUES IN CELL BIOLOGY
2. 5.2 FRAP EXPERIMENTS
3. 5.3 FRAP DATA ANALYSIS
4. 5.4 PROCEDURES FOR QUANTITATIVE FRAP ANALYSIS WITH FREEWARE SOFTWARE TOOLS
5. 5.5 NOTES
6. 5.6 CONCLUDING REMARKS
7. 5.7 REFERENCES
6 Co‐Localisation and Correlation in Fluorescence Microscopy Data
1. 6.1 INTRODUCTION
2. 6.2 CO‐LOCALISATION FOR CONVENTIONAL MICROSCOPY IMAGES
3. 6.3 CONCLUSION
4. 6.4 ACKNOWLEDGMENTS
5. 6.5 REFERENCES
7 Live Cell Imaging
1. 7.1 INTRODUCTION
2. 7.2 SETTING UP A MOVIE FOR TIME‐LAPSE IMAGING
3. 7.3 OVERVIEW OF AUTOMATED AND MANUAL CELL TRACKING SOFTWARE
4. 7.4 INSTRUCTIONS FOR USING IMAGEJ TRACKING
5. 7.5 POST‐TRACKING ANALYSIS USING THE DUNN MATHEMATICA SOFTWARE
6. 7.6 SUMMARY AND FUTURE DIRECTION
7. 7.7 REFERENCES
8 Super‐Resolution Data Analysis
1. 8.1 INTRODUCTION TO SUPER‐RESOLUTION MICROSCOPY
2. 8.2 PROCESSING STRUCTURED ILLUMINATION MICROSCOPY DATA
3. 8.3 QUANTIFYING SINGLE MOLECULE LOCALISATION MICROSCOPY DATA
4. 8.4 RECONSTRUCTION SUMMARY
5. 8.5 IMAGE ANALYSIS ON LOCALISATION DATA
6. 8.6 SUMMARY AND AVAILABLE TOOLS
7. 8.7 REFERENCES
9 Big Data and Bio‐Image Informatics
1. 9.1 INTRODUCTION
2. 9.2 WHAT IS BIG DATA ANYWAY?
3. 9.3 THE OPEN‐SOURCE BIOIMAGE INFORMATICS COMMUNITY
4. 9.4 COMMERCIAL SOLUTIONS FOR BIOIMAGE INFORMATICS
5. 9.5 SUMMARY
6. 9.6 ACKNOWLEDGMENTS
7. 9.7 REFERENCES
10 Presenting and storing data for publication
1. 10.1 HOW TO MAKE SCIENTIFIC FIGURES
2. 10.2 PRESENTING, DOCUMENTING AND STORING BIOIMAGE DATA
3. 10.3 SUMMARY
4. 10.4 REFERENCES
11 Epilogue
1. 11.1 WORKFLOWS FOR BIOIMAGE ANALYSIS
2. 11.2 RESOURCES FOR DESIGNING WORKFLOWS AND SUPPORTING BIOIMAGE ANALYSIS
3. 11.3 CONCLUSION
4. 11.4 REFERENCES
Index
End User License Agreement

List of Tables

Chapter 01
1. Table 1.1 List software for measuring and processing bioimages.
Chapter 02
1. Table 2.1 Standard object intensity descriptors.
2. Table 2.2 Summary of some object shape descriptors in 2D and 3D.
Chapter 03
1. Table 3.1 Commonly used Bioimaging software for quantification.
Chapter 05
1. Table 5.1 Software tools for FRAP data processing.
Chapter 08
1. Table 8.1 SIM data quality checks (SIMcheck).
2. Table 8.2 Examples of SMLM analysis software and its features. This non‐exhaustive list provides an overview of existing software and the included characteristics. Note: for 3B and FALCON, some grouping is performed by default by the software during the peak localisation step (‘implicit’) but grouping cannot be done separately.
Chapter 11
1. Table 11.1 Types of components.

List of Illustrations

Chapter 01
1. Figure 1.1 Bioimage quantification to determine the dynamics of actin using photoconversion. Tsang, Wheeler and Wan Experimental Cell Research, vol. 318, no. 18, 01.11.2012, p. 2269–83.
2. Figure 1.2 Workflow for bioimage data capture in 2D and 3D.
3. Figure 1.3 Combining channels in fluorescent bioimage analysis. Channel 1 has antibodies raised against E‐cadherin labelled with AlexaFluor 568 secondary antibodies. Channel 2 is labelled with primary antibodies raised against Alpha tubulin and secondary antibodies labelled with AlexaFluor 488.
4. Figure 1.4 The Bioimage analysis workflow.
5. Figure 1.5 How images are digitised.
6. Figure 1.6 Basic quantification of cellular features using 8‐bit fluorescent image of F‐actin.
7. Figure 1.7 The effect of saturation and under‐sampling on bioimage analysis.
8. Figure 1.8 Binning of pixels to increase speed and sensitivity of Bioimage acquisition.
9. Figure 1.9 Bucket brigade CCD analogy
10. Figure 1.10 A 3 × 3 median filter kernel. The filter size is indicated in orange. This filter smooths the image and denoises it.
11. Figure 1.11 Initialisation using filtering (a) Illustrative example of image filtering taken from the Image J webpage https://www.fiji.sc, (b) Example of rolling ball background subtraction: left‐hand side is before correction, and right‐hand side after, (c) Using ROI subtraction.
12. Figure 1.12 Experimental point spread functions: By Howard Vindin (own work) [CC BY‐SA 4.0 (http://creativecommons.org/licenses/by‐sa/4.0)], via Wikimedia Commons.
13. Figure 1.13 Using ImageJ to select parameters of shape and intensity in an image of nuclei (blue). Here nuclei have been manually segmented using a contour – yellow line. Measurements have been set in ImageJ, and the numerical results output. The area of each nucleus, mean, standard deviation, maximal and minimal intensity are computed. The circularity (Circ) of the nuclei are also computed.
Chapter 02
1. Figure 2.1 (a) Magritte, “The treachery of Images” extract, Los Angeles County Museum of Art. (b) Cells stained in immune‐fluorescence. Reproduced with kind permission of Design and Artists Copyright Society, UK.
2. Figure 2.2 (a–c) GFP‐NEMO–expressing fibroblasts. Upon treatment with IL‐1, the soluble protein aggregates in small punctate, transient structures that can be detected by live‐cell fluorescence microscopy, and disappear 10 minutes after stimulation. Scale bar: 10 µm. Panels a, b and c show the first, 20^th and 40^th frame respectively of the movie post‐stimulation. (d) Mean intensity and its standard deviation calculated inside the yellow ROI in preceding panels, normalized with their value at frame 1. Sample courtesy of Nadine Tarantino and Emmanuel Laplantine, Institut Pasteur, Paris, France.
3. Figure 2.3 (a) Simulation of 2000 particles spread over 200 × 200 µm. The particules were spread uniformly, then their position was updated by a quantity equal to towards the center with r being their distance to the center. In effect, this moves mainly the particles in a radius of 25 µm towards the center, creating a small aggregation below this radius and depletion at this radius. Beyond 50 µm away from the center, the particle positions are left unchanged, and their distribution remains random. (b) The Ripley’s K function for this particle distribution (blue) and for a similar random distribution (black). Dashed red lines: 95% confidence interval for complete spatial randomness assessed by numerical simulations. (c) The Besag L function for this distribution. The function peaks around 60 µm, giving a broad estimate of the aggregate size.
Chapter 03
1. Figure 3.1 (a) L929 cell observed in bright‐field. A pulsed 405 nm laser was used to ablate the cell cortex in its periphery, which generated a bleb that can be see growing on the top‐left part of the cell. (b) The cell image, overlaid with the result of segmentation (red) and the two‐circle fit (yellow). (c) Two cells aspired in a micropipette, observed in bright‐field. (d) The cells image overlaid with the two‐circle fit (yellow).
2. Figure 3.2 (a) L929 cell observed in bright‐field overlaid with the segmentation results of the image in b. (b) Image from a. transformed with a 3 × 3 standard deviation filter. Scale bar: 5 µm.
3. Figure 3.3 (a) Macrophages fixed and stained with DAPI observed in fluorescence. (b) Close up version of the red box in a. The green line delineates a contour obtained by putting a threshold on intensity at 5000. (c) Binary mask obtained by thresholding intensity at 5000. Note that the masks of the two cells indicated by the red arrow are touching. (d) Objects after labeling. Different objects are indicated by different colors. The two touching cells generated a single object. Sample courtesy of Thibault Rosazza and Eric Prina, Institut Pasteur, Paris, France.
4. Figure 3.4 Some morphological operations. (a) Synthetic example mask generated by taking portions of figure 4a. (b), (c), (d), (e), (f). Results of respectively the dilation, erosion, watershed, close and open operations on the source image in a. Dilation, erosion, close and open operations used a 3 × 3 square structuring element.
5. Figure 3.5 (a) Epithelium of the neural plate of Xenopus laevis visualized by utrophin‐GFP. (b) Results of segmentation using the Morphological Segmentation Fiji plugin. Image courtesy of Jakub Sedzinski, UT Austin, USA.
6. Figure 3.6 The watershed segmentation technique. (a) Synthetic noisy image depicting two compartments delineated by bright and noisy contours. (b) Watershed technique illustrated on an intensity profile through the red line in a. The profile was filtered to appear smooth. From left to right the “water level” rises, flooding compartments. The object contours are set where waters from two basins meet. (c) Actual watershed on a non‐filtered profile. Intensity fluctuations due to noise generate several small basins that lead to numerous spurious objects. (d) Marker‐controlled watershed. Water flows only from a discrete numbers of sources set by markers. As the water level rises, it floods the small, noisy basins. In the end, this technique generates one object per marker.
7. Figure 3.7 (a) Epithelium of the neural plate of Xenopus laevis visualized by utrophin‐GFP. Scale bar: 10 µm. (b) Segmentation results using the Morphological Segmentation Fiji plugin. (c) Initialization contour over a target cell. Note that the top left membrane of the target cell displays a hole in its staining. (d) Results of segmentation with the Active Contour plugin of Icy. The technique could successfully bridge over the membrane gap. Image courtesy of Jakub Sedzinski, UT Austin, USA.
8. Figure 3.8 (a) Transmitted electron micrographs of the basolateral part of enterocyte cells in the guinea pig colon. The color overlay shows manual annotation of three parts of the cells: mitochondria (red), cytoplasm (green) and basolateral plasma membrane (purple). (b) Results of the supervised machine learning segmentation, using the Trainable Weka Segmentation plugin and the input classification depicted in panel a. Image courtesy of M. Sachse and E.T. Arena, Institut Pasteur, Paris, France.
Chapter 04
1. Figure 4.1 FRET phenomenon is used to measure association kinetics of biomolecules. Sample is illuminated in the donor excitation spectral range. If donor and acceptor molecules are far away from each other (left) emission is detected only in the donor channel. When the two fluorophores get closer (right), FRET can occur, resulting in decrease of donor emission and increase of acceptor emission.
2. Figure 4.2 Measuring intracellular calcium levels using FRET‐based sensors and ratiometric imaging.
3. Figure 4.3 Measuring FRET by acceptor photobleaching.
Chapter 05
1. Figure 5.1 FRAP measures mobility of biomolecules. Steady‐state spatial distribution of labelled molecules is detected from prebleach images. Bleaching of the selected ROI perturbs equilibrium of fluorescence distribution, which is restored over time due to replacement of the bleached fluorophores in the ROI with non‐bleached fluorophores. FRAP recovery is quantified as an average or total intensity in the bleached ROI.
2. Figure 5.2 Use of FRAP to measure reaction‐diffusion kinetics of molecules in living cells
3. Figure 5.3 Workflow for quantitative analysis of FRAP recovery curves
4. Figure 5.4 Quantification of fluorescence intensity traces in time‐lapse images with ImageJ/Fiji
5. Figure 5.5 FRAP with COPII components at ER‐exit sites
Chapter 06
1. Figure 6.1 Schematic intensity scatter plots between two fluorescence channel intensities for the case of (a) perfect co‐localisation, (b) partial co‐localisation, (c) partial exclusion and (d) total exclusion.
2. Figure 6.2 Schematic of Van Steensel’s method. If two fluorophores (red and green in this example) do not show complete co‐localisation but do show a relation, calculation of Pearson’s correlation coefficient, R, in the presence of shifts in the position of one image relative to the other (in this case along the x axis) will display peaks, with the position of the peaks indicating the spatial separation of the fluorophores.
3. Figure 6.3 (a) Method for co‐cluster analysis using Getis’ variant of Ripley’s K‐function. Red molecules are counted within a radius centred on each green molecule, and vice versa, to give the L(r)cross values (in this case the radius = 50). The L(r) values are then plotted against L(r)cross for each channel. The resulting plots can then be divided into quadrants and, by applying thresholds, the strength of the correlation can be determined to assess co‐clustering. Colours indicate how many molecules in a region of interest had that specific combination of L(r) and L(r)cross value.
4. Figure 6.4 Basic setup and principle of fluorescence correlation spectroscopy. Excitation light is directed through the high NA objective into the sample, creating the excitation volume. Fluorescent molecules travelling into this volume are excited and their fluorescence is collected back through the objective onto a photon detector.
5. Figure 6.5 By correlating a data set with itself over varying time lags (τ), underlying characteristics can be extracted. Intensities measured at shorter time lags (τ1) tend to show similarities giving a higher correlation. Over longer time lags (autocorrelation τ2) the data is less similar, resulting in a reduced correlation. Once the correlation function (G(τ)) is plotted over time, the point at half the maximum amplitude gives the correlation time τ_D.
Chapter 07
1. Figure 7.1 Display window for ImageJ running on Windows 7 Professional.
2. Figure 7.2 Interface for manual tracking plugin opened in ImageJ.
3. Figure 7.3 Example of the drawing features available in manual tracking to allow the user to identify which cells have been tracked.
4. Figure 7.4 Window displaying the results of cell trajectories gained during tracking using the Manual Tracking plugin.
5. Figure 7.5 Wolfram Mathematica 7.0 interface displaying the chemotaxis notebook program.
6. Figure 7.6 Initiation message that appears over the chemotaxis notebook when the user presses Shift Enter.
7. Figure 7.7 Importing tracking files into Mathematica.
8. Figure 7.8 Track plotting, analysing and editing window as displayed in Mathematica.
9. Figure 7.9 Two sample t‐test window as displayed in the Mathematica chemotaxis notebooks.
10. Figure 7.10 Rayleigh test of Directions window as displayed in the Mathematica chemotaxis notebooks.
11. Figure 7.11 Moore test on ranked vectors window as displayed in the Mathematica chemotaxis notebooks.
Chapter 08
1. Figure 8.1 Structured Illumination increases resolution by extending the effective observable frequency range. Structured illumination generates frequencies that the imaging system can capture by frequency mixing between the striped (sinusoidal) illumination pattern and subresolution structures in the sample: these are analogous to the Moiré fringes (large vertical stripes) shown in (a). For the 2D case considered here (b), the illumination pattern contains three Fourier components. The central zero order component corresponds to frequencies observable in a standard wide‐field image. Three illumination pattern phases are required to separate the three components shown in black, where the outer circles represent the region of the frequency space sampled by first‐order components. The red circles represent two additional angles, which are required for isotropic sampling of all xy frequencies.
2. Figure 8.2 SIM data quality control. (a) SIMcheck is an ImageJ plugin suite designed to run a series of checks on raw and reconstructed SIM data and produce a report summarising the results. (b–e) SIM data are from Molecular Probes FluoCells Slide #1 (BPAE cells stained with DAPI, Alexa Fluor 488 Phalloidin and MitoTracker Red), acquired by Paul Appleton, on a GE OMX V4 Blaze (Dundee Imaging Facility): (b) shows a projection of Fourier transformed raw data channel 1 (DAPI) with the zero order spot blanked out, first‐ and second‐order spots clearly visible, (c) shows average (slice) intensity profiles for the raw data as phase, z position and angle are incremented, (d) shows a 2D Fourier transform of the channel 2 (phalloidin) reconstruction, with ‘resolution rings’ in microns, (e) shows the channel 2 reconstruction colour‐coded according to modulation contrast‐to‐noise (mostly orange‐yellow, indicating >10, which is good).
3. Figure 8.3 Sequence of steps from raw SMLM images to reconstructed super‐resolved image. Examples of software as well as typical sample type are indicated for each step. Typically 10,000 raw images containing single molecule events are acquired, followed by (1) localisation of the centre of each molecule of each frame through a fitting process resulting in a list of localisations, (2) subsequent sieving of this list discards noise from ‘real peaks’ to generate a curated list of ~100,000 localisations, (3) additional corrections such as de‐drifting or grouping of multiple emitters (see also Figure 8.4) are performed before (4) rendering of a super‐resolved (SR) image or (5) quantification of specific features, e.g. structure size, cluster organisation or stoichiometry.
4. Figure 8.4 Schematic of the ‘grouping’ or ‘blink‐correction’ procedure. (a) Temporal and (b) spatial intensity traces of a small image region showing multiple emission peaks over several frames, characteristic of a single molecule blinking behaviour. All the localisations within a given region of space limited by a grouping radius r_g (b), and time limited by a grouping time τ_g (a) are considered to originate from the same molecule and their positions are averaged into a new grouped photon‐weighted average molecular localisation.
Chapter 09
1. Figure 9.1 User interface of ImageJ and Fiji.
2. Figure 9.2 Fiji macro scripting console.
3. Figure 9.3 Batch Process tool in Fiji.
4. Figure 9.4 User interface of CellProfiler. It is possible to rapidly query images of interest using keywords.
5. Figure 9.5 A typical cluster configuration.
6. Figure 9.6 Linux HPC where the user has accessed a log‐in node of a cluster.
7. Figure 9.7 User interface of Icy displaying a three‐dimensional dataset.
8. Figure 9.8 User interface of Imaris displaying a three‐dimensional dataset.
9. Figure 9.9 Example of how Definiens could be used to rapidly classify different types of cells on histology images. On the left is the original scan, and the rightimage shows how Definiens has classified the cells into two classes.
Chapter 10
1. Figure 10.1 representative example of a scientific figure which is acceptable for publication. A super‐resolution (STORM) image of a HeLa cell nucleus labelled with antibodies to HP1 alpha (red) and LaminB (green). Scale bar indicated 2 microns. The high resolution insert, right, is indicated by a yellow box on the merged panel.
2. Figure 10.2 Using Fiji/ImageJ to open data using the Bio‐Formats importer.
3. Figure 10.3 The Bio‐Formats importer dialogue box.
4. Figure 10.4 A schematic diagram of a storage area network https://en.wikipedia.org/wiki/Storage_area_network.
5. Figure 10.5 The OMERO server enables storage, annotation and analysis of image data in one easy‐to‐use interface.
Chapter 11
1. Figure 11.1 Four types of bioimage analysis workflow. The left column shows the schematic definition of component types (also refer to Table 11.1). The right column shows four type of workflows as flowcharts: Type 0, the simplest workflow segmenting a single type of object from a single channel; Type 1, single‐ or dual‐channel dual‐branch workflow annotating a specific object type and then measuring intensity inside these objects in the raw images; Type 2, dual‐channel dual‐branch co‐measuring two object types; Type 3, dual‐channel dual‐branch detecting seeds in one channel to segment objects in the other channel. Original design: Romain Guillet.
2. Figure 11.2 Integrated workflow chart of bioimage analysis. Each box is a component type, and the colours correspond to the three conceptual types defined in Table 11.1. Pink and grey: image processing components; Blue and orange: image analysis components; Blue‐grey and purple: data analysis components.
3. Figure 11.3 Results of a survey in 2015. To the question ‘Which resource do you miss the most for optimally analysing images?’, 67.8% of respondents, mostly from the bioimaging community, chose either support from specialists or courses (n = 1904). The colors are for different expertise identified by each of the responders by themselves.

This edition first published 2018
© 2018 John Wiley & Sons Ltd

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by law. Advice on how to obtain permission to reuse material from this title is available at http://www.wiley.com/go/permissions.

The right of Ann Wheeler and Ricardo Henriques to be identified as the authors of the editorial material in this work has been asserted in accordance with law.

Registered Office(s)
John Wiley & Sons, Inc., 111 River Street, Hoboken, NJ 07030, USA
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

Editorial Office
The Atrium, Southern Gate, Chichester, West Sussex, PO19 8SQ, UK

For details of our global editorial offices, customer services, and more information about Wiley products visit us at www.wiley.com.

Wiley also publishes its books in a variety of electronic formats and by print‐on‐demand. Some content that appears in standard print versions of this book may not be available in other formats.

Limit of Liability/Disclaimer of Warranty
In view of ongoing research, equipment modifications, changes in governmental regulations, and the constant flow of information relating to the use of experimental reagents, equipment, and devices, the reader is urged to review and evaluate the information provided in the package insert or instructions for each chemical, piece of equipment, reagent, or device for, among other things, any changes in the instructions or indication of usage and for added warnings and precautions. While the publisher and authors have used their best efforts in preparing this work, they make no representations or warranties with respect to the accuracy or completeness of the contents of this work and specifically disclaim all warranties, including without limitation any implied warranties of merchantability or fitness for a particular purpose. No warranty may be created or extended by sales representatives, written sales materials or promotional statements for this work. The fact that an organization, website, or product is referred to in this work as a citation and/or potential source of further information does not mean that the publisher and authors endorse the information or services the organization, website, or product may provide or recommendations it may make. This work is sold with the understanding that the publisher is not engaged in rendering professional services. The advice and strategies contained herein may not be suitable for your situation. You should consult with a specialist where appropriate. Further, readers should be aware that websites listed in this work may have changed or disappeared between when this work was written and when it is read. Neither the publisher nor authors shall be liable for any loss of profit or any other commercial damages, including but not limited to special, incidental, consequential, or other damages.

Library of Congress Cataloging‐in‐Publication Data

Names: Wheeler, Ann, 1977– editor. | Henriques, Ricardo, 1980– editor.
Title: Standard and Super‐Resolution Bioimaging Data Analysis: A Primer / edited by Dr. Ann Wheeler, Dr. Ricardo Henriques.
Description: First edition. | Hoboken, NJ : John Wiley & Sons, 2018. | Includes index. |
Identifiers: LCCN 2017018827 (print) | LCCN 2017040983 (ebook) | ISBN 9781119096924 (pdf) | ISBN 9781119096931 (epub) | ISBN 9781119096900 (cloth)
Subjects: LCSH: Imaging systems in biology. | Image analysis–Data processing. | Diagnostic imaging–Data processing.
Classification: LCC R857.O6 (ebook) | LCC R857.O6 S73 2017 (print) | DDC 616.07/54–dc23LC record available at https://lccn.loc.gov/2017018827

Cover design by Wiley
Cover image: Courtesy of Ricardo Henriques and Siân Culley at University College London

Foreword

Imaging is now one of the most commonly used techniques in biological research. It is not simply a means of taking a pretty picture; rather advanced microscopy and imaging are now vital biophysical tools underpinning our studies of the most complex biological systems. We have the capability to study cells in real time, with 3D volumes, analysing biophysical interactions, and now moving towards the ability to see and study individual proteins within individual cells within their native environment. Imaging is one of the most useful tools for understanding the fundamental biology of the cell.

This has been made possible through an incredibly rapid period of microscopy development, which has gone hand in hand with the emergence of new fluorescent tags – such as fluorescent proteins – computing power and the latest developments in engineering. Not only has the technology become increasingly versatile and opened up many new possibilities, but leading manufacturers have also made their microscopes increasingly accessible to non‐specialists, resulting in an explosion of data.

All of these developments have left us with a wealth of data, but the images themselves will remain just pretty pictures unless they are analysed appropriately. We are now starting to see an equivalent rapid increase in the development of image analysis, but we are still far from realising its full potential. The basics are vital, and anyone using today’s microscopes should also be looking at the best approaches for analysing their data.

This book is extremely timely and looks at some of the key aspects of data analysis; it will serve as an excellent point of reference. Chapter 1 examines the basics of image data and processing which is common to most users. Chapters 2 and 3 builds on this and looks at how to quantify both routine 2D through to more complex 3D image data sets. However, one of the greatest challenges remains the ability to segment our images. To our own eyes, this can be quite obvious, but it still remains a real computational challenge. Segmenting images and image data e.g. in Chapter 3 will be a continuing area of development that will also help us to correlate data with greater precision in the future.

Beyond the images, the microscope is a powerful biophysical tool. We can see and analyse the co‐localisation of particles such as proteins, but great care is needed in their analyses as outlined by Dylan Owen e.g. in Chapter 6. We can now go beyond this and start to see interactions that occur over a 5 nm range. This enables the studies of particle–particle interactions such as protein–protein heterodimerisation by using FRET, and although the imaging of these phenomena is relative simple the complexity of the quantification is discussed in Chapter 4.

Not only can the microscope study these natural interactions, but we can also use the light, often with lasers, to manipulate cells and trigger critical events that would otherwise occur in a random fashion making their studies very difficult. The interpretation and controls for FRAP and other photo perturbation methods then needs to be carefully considered (Chapter 5).

Many of the above studies can be undertaken using both fixed and live cell imaging. Whole live‐cell imaging and tracking brings its own analytical challenges (Chapter 7) as cells move through three dimensions, pass over one another often changing shape and nature. This needs many of the above elements to come together to help work in such complex sample types.

At the other extreme, from whole cells, the biggest advancement in light microscopy has come from the ability to image below the diffraction limit. Super‐resolution microscopy (SRM) is possible through many different strategies. The analysis and interpretation is an area that is often under‐appreciated and which can result in misinterpretations. For anyone wanting to undertake SRM, it is essential to understand the limitations, controls and best approaches to their analysis (Chapter 8).

Many of the new microscopical techniques now produce very large data sets. This is especially true for 3D live‐cell imaging and SRM. This has developed its own problem, with data analyses now often taking considerably longer than the imaging time itself. The time needed for data analysis has become the most costly element in complex image study. The staff time itself often outweighs the cost of the instrument and consumables and this is why we need to look for automation when handling and analysing the data (Chapter 9), and naturally, once all of this data has been analysed, it is vital to not only present the data in the correct manner, but also to ensure that it is correctly documented and stored (Chapter 10). Only then, can any one image be properly exploited and deliver the required impact.

Peter O’Toole
York
July 2017