Presenting data and making figures BALM facility talk 2014

Presenting data and making figures
Isma Ali
Blizard Advanced Light Microscopy
Representative
labels
Colour image,
in focus
Clear staining
Inserts
Images
presented
regularly in a
montage
No background
noise or
spectral bleed
through
Clear figure legend
Scale bars
Figure 8. Laminin staining is disrupted at the bridges between acini. On day 4, cultures of MCF10A acini were fed in the A) absence or B) and
C) presence of 10 nM TPA. The cultures were fixed 23 hours later and stained with DAPI to detect nuclei (blue) and either phalloidin to
detect actin (green, left panels) or an anti-laminin antibody (red, right panels) (40X). Bar = 50 microns. Inserts in B) show regions that were
digitally enlarged to show areas of the bridge (in boxes) that have irregular laminin staining (red, right panel); actin staining for the same
region is also shown (green, left panels). Arrows in C) show disrupted laminin staining (red, right panel). Representative images are shown.
Klos KS, Warmka JK, Drachenberg DM, Chang L, Luxton GWG, et al. (2014) Building Bridges toward Invasion: Tumor Promoter Treatment
Induces a Novel Protein Kinase C-Dependent Phenotype in MCF10A Mammary Cell Acini. PLoS ONE 9(3): e90722.
doi:10.1371/journal.pone.0090722
Best practise in data presentation
Human lung fibroblast showing intracellular proteins.
Epi-fluorescence microscopy, 20x objective. A. Singh,
Cornell University, United States. Scale bar: 40 microns.
Figure 5 | Effects of 1800 MHz RF-EMF exposure on eNSC
differentiation. (a) The representative images of Tuj11 and GFAP1
cells. Scale bar: 50 microns. (b) Statistical results of the percentage
of Tuj11 and GFAP1cells in each condition. Laminin staining
disruption at bridges between acini, 40x objective with digitally
enlarged insets. C. Chen et al., Department of Occupational
Health, Faculty of Preventive Medicine, Third Military Medical
University, No.30 Gaotanyan Street, Chongqing 400038, China.
Points we are covering
•
•
•
•
•
•
•
Microscope light path
How an intensity camera works
Adding colour to greyscale images
Scale bars
Background subtraction
Stacks and montages
Making an insert
Microscope light path
Light from light source
illuminates specimen
Light moves through
microscope
Detected on camera chip
Recorded on an array of
pixels
How an intensity camera works
• Camera only takes grey images, as pixels can detect any colour
• When light falls on pixels, it is given a numeric value based on it’s intensity
• Number of greyscales camera can detect, determines numeric range (e.g.
12 bit = 4,096 greyscales)
Camera
Specimen
13
45
23
9
67
243
78
55
23
15
210
70
61
3
167
20
Light hits silicon semi-conductor and emits
electrons = Photo-electric effect = Signal
How an image is formed
•
•
Scientific camera is formed of an array of pixels of specific dimensions
Save images as TIFFs to preserve raw data, JPEGs compress the image
A CCD chip has an array of small
photosensitive elements
Real image
75
75
75
75
75
75
75
65
52
84
78
72
255
153
68
87
68
65
72
85
81
206
168
43
67
57
65
68
59
85
255
255
20
106
40
65
33
65
102
240
235
178
22
78
65
0
28
52
176
200
185
95
58
55
0
0
1
3
2
4
3
5
2
Individual pixels
An image is a numeric matrix of pixels
What is bit depth?
Bit depth is the intensity range expected from each pixel. Greyscale values can be
read and analysed by a computer.
Higher bit depth
Better resolution of image
Which bit depth should I use?
Qualitative
images, keeps
colour
Quantification,
grey images
High resolution,
grey images
Bit Depth
Grayscale
Levels
8
9
10
256
512
1,024
11
2,048
12
13
14
16
4,096
8,192
16,384
65,536
• 8 bit = 256 discrete greyscale levels (between 0 and 255). Ideal for
qualitative images, keeps image colour.
• 12 bit = 4,096 greyscale levels. Ideal for quantification, images are grey yet
colour can be added back.
Duplicating image
It is useful to duplicate before processing, so you don’t loose the original image:
Or use Ctrl + Shift + D
Save duplicate!
How to add colour back to images
•
•
•
12/16 bit images appear grey, as saved in greyscale colour format.
8 bit format keeps image colour, but less depth of intensity information.
Colour can easily be added back:
Please see protocol for putting colour back into greyscale images:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/How%20to%20put%20the%20colour%20back%2
0into%20grey%20images.pdf
Image background
High background image
High background
Weak staining
Strange artefacts
Background subtracted image
Background removed
How to subtract background
1) Removing uneven illumination, out of focus light and autofluorescence:
Rolling Ball background correction evens out uneven signal.
After background subtraction
Before
Same size as the smallest object in image
2) To get rid of ‘constant’ noise:
1
Draw region with
only background
staining
Measure average background intensity
2
Please see protocol for Background subtraction:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Background%20subtraction.pdf
How to subtract background
4
3
Enter the background intensity in the Value box:
5
This intensity value will
be subtracted from all
pixels in the image,
removing detector noise
3) Or you can use the background subtraction plugin:
Default = 2.0
It then subtracts from the
image: [mean + (sd × scaling
factor)]
Please see protocol for Background subtraction:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Background%20subtraction.pdf
Denoising images
Residual noise in images can be caused by excess antibody staining, cellular debris or
noise from the detector:
Despeckle replaces each pixel with the median value
Remove Outliers replaces a pixel by the median of pixels if it
deviates from median by more than the threshold
Median filters: used to eliminate camera noise or PMT noise
Please see protocol for Background subtraction:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Background%20subtraction.pdf
Contrast stretching
•
Contrast stretching can help to better
highlight data present
•
Image, Adjust, Brightness/Contrast
(Ctrl + Shift + C)
•
Alter the minimum, maximum, brightness
or contrast, and press Apply
If you have a stack of images, it
will alter the brightness and
contrast of the whole stack at
the same time!
How to make a stack
•
•
Stacks of images can be useful for batch processing
You can perform the same operation on all images in a stack, e.g. resizing, changing
colour and measuring
1. Open the images you want to use
2. Image, Stacks, and Images to Stack
Input a name for the stack, and type in a word or
abbreviation that appears in all the image names
Please see protocol for making stacks and montages:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Stacks%20and%20montages.pdf
Stacks for large data sets
•
•
•
Scroll to move through stack:
For importing lots/selection of images, (e.g. live
cell imaging):
In Sequence Options window, tell ImageJ which
images to open
3
1
•
•
•
All images must be
the same size and file
format
Enter common word
to open images
Save stack!
2
Please see protocol for making stacks and montages:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Stacks%20and%20montages.pdf
Sorting stacks
To move images around in a stack and sort:
Check images are in correct order, stack
position and image name
Plugins, Stacks - Shuffling and
Stack sorter
Please see protocol for making stacks and montages:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Stacks%20and%20montages.pdf
Example of a montage
Confocal images showing localisation of GSDMB-HA isoforms (green) and RAC-1 (red) in cancer cells. Scale
bar = 37.5 microns.
Hergueta-Redondo M, Sarrio´ D, Molina-Crespo A´, Megias D, Mota A, et al. (2014) Gasdermin-B Promotes
Invasion and Metastasis in Breast Cancer Cells. PLoS ONE 9(3): e90099. doi:10.1371/journal.pone.0090099
How to make a montage
•
•
•
Allows presentation of images together in ordered sequence
Easy to compare differences or changes in images
Open a stack with the images you want. Then go to Image, Stacks, and Make Montage:
•
•
•
Usually number
columns =
number epitopes
stained
Number rows =
number of
conditions
Border width: at
least 5
Save montage!
To see more detail about montages see protocol:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Stacks%20and%20montages.pdf
Calibrating your image
• Measurements in images are originally made in pixels
• Calibrate image first, so it is converted from pixels to microns (pixel to
micron ratio)
• Magnification of objective and camera pixel size determines the scale
ratio found in calibration information on Z drive:
What is a scale bar?
• Shows the scale of area of interest in the image, such as size of a cell
• Add scale bar size information in figure legend
• Calibration information on Z drive and scale bar protocol available
Scale bars
Scale bar: 60 microns
Scale bar: 50 microns
How to add a scale bar
1
2
3
4
5
Please see scale bar protocol for adding scale bars using Zen and Stereoinvestigator:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Scale%20Bars.pdf
Don’t forge images!
• It is important not alter the image in a way that misrepresents
the true image, e.g. with misrepresentative inserts
Original image
Insert added from
another image,
misrepresenting
true image
Inaccurately altered image
Example of inserts in figures
H&E staining of tumour cells at x20 magnification and with
x60 magnification insets. Hergueta-Redondo M, Sarrio´ D,
Molina-Crespo A´, Megias D, Mota A, et al. (2014)
Gasdermin-B Promotes Invasion and Metastasis in Breast
Cancer Cells. PLoS ONE 9(3): e90099.
doi:10.1371/journal.pone.0090099
Confocal images of fixed FAK-/- cells with insets showing focal
adhesions. Deramaudt TB, Dujardin D, Noulet F, Martin S,
Vauchelles R, et al. (2014) Altering FAK-Paxillin Interactions
Reduces Adhesion, Migration and Invasion Processes. PLoS ONE
9(3): e92059. doi:10.1371/journal.pone.0092059
How to make an insert
•
Inserts can be added to an image to
highlight a point, as long as they are not
interfering with the true image.
•
•
Resizing changes the scale of images.
Cropping keeps the scale the same.
Confocal images of FAK-/- cells
showing Src and paxillin
staining with magnified insets.
Deramaudt TB, Dujardin D,
Noulet F, Martin S, Vauchelles
R, et al. (2014) Altering FAKPaxillin Interactions Reduces
Adhesion, Migration and
Invasion Processes. PLoS ONE
9(3):e92059.
doi:10.1371/journal.pone.009
2059
Inserts
showing
magnified
regions
Resizing
Cropping
Cropping images
To keep the scale of the image, areas can be cropped and added as inserts:
To crop, use the selection
tools and highlight the
area you want to keep,
then go to Image, Crop:
Please see resizing and cropping protocol:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Resizing%20or%20cropping%20your%20images.pdf
Resizing images
Details of the size of your image
Go to Image, Adjust, Size
Type in the size you want
your image to be (in pixels)
New image size
Please see resizing and cropping protocol:
http://www.icms.qmul.ac.uk/imaging/Protocols/Image%20analysis/Resizing%20or%20cropping%20your%20images.pdf
Protocols we discussed today
BALM Image analysis protocols:
http://www.icms.qmul.ac.uk/imaging/Image%20Analysis.html
Protocols discussed today:
• Putting colour back to greyscale images
• Adding a scale bar to your image
• Background subtraction
• Making stacks and montages
• Resizing or cropping your images
• How to generate a scientific figure
Further resources
• Olympus Microscopy Resource Centre
http://www.olympusmicro.com/index.html
• Nikon MicroscopyU http://www.microscopyu.com
• Fluorescence SpectraViewer
• http://www.lifetechnologies.com/uk/en/home/lifescience/cell-analysis/labeling-chemistry/fluorescencespectraviewer.html
For more help, please contact me!
[email protected], Ext: 2407