Post on 05-Jul-2020
Fundamentals of Digital Imaging
Dr Paul McMillan
Biological Optical Microscopy Platform
1
Fundamentals of digital imaging• The Digital Image (pixels, bit depth)• Image Acquisition (dynamic range, resolution, sampling,)• Basic Rules of Digital Images (data handling & image ethics)• Image Presentation (guidelines, colour sensitivity/blindness
& printing)
Applications (Demonstrations & hands on sessions)• Image Import• Image manipulation• Annotation• Visualisation & presentation of 2D data• Visualisation & presentation of 3D data• Visualisation & presentation of timelapse data
FIJI/Image J for Beginners
The Digital Image
Dr Paul McMillan
Biological Optical Microscopy Platform
Pixels
Your image is recorded as a numerical array of pixels
Pixel has its defined - size (sampling resolution)- grey level (bit depth)
100nm
100nm
Bit depth
1 bit2 grey levels
21
2 bit4 grey levels
22
3 bit8 grey levels
23
4 bit16 grey levels
24
Bit depth – intensity coding
8bit
175 175
16bit
44679 44830
Bit depth – Example images
1-bit2-bit3-bit4-bit8-bit
Bit depth
0
50
100
150
200
250
1-bit
2 bit
3-bit
4-bit
8-bit
Bit depth - equipment
Bit depth Grey levels Use BOMP equipment
1 2 Binary masks for image analysis
FIJI, Imaris, Metamorph, Volocity
5.65 50 Late night reading NA
8 256 Scientific cameras, PMT, GaAsP
PMT, GaAsP (LSM800)
12 4096 Scientific cameras, PMT, HyD PMT, HyD (Leica SP8)
15 32,768 Scientific cameras sCMOS (OMX)
16 65,536 Scientific cameras, PMT, GaAsP
PMT, GaAsP (LSM800)
32 4,294,967,296 FLIM & other high end techniques
Visualisation = 8 bit, quantification = 12 or 16 bit
Image acquisition
Dr Paul McMillan
Biological Optical Microscopy Platform
Intensity graph
Intensity distribution of image
x axis = intensity
Y axis = # of pixels with that intensity
Intensity graph
All image modification must be ETHICAL
Can adjust Minimum & Maximum position
DO NOT adjust gamma (non-linear)
Be descriptive of modifications in methods
VECTOR RASTER
Resolution & Sampling
10x10
0.1
Frame size3x3
Pixel size 1/3 = 0.3
5x5
0.2
20x20
0.05
300x300
0.003
Resolution & Sampling
Nyquist rate
In order to convert analogue signal to digital, signal needs to be sampled at least 2.3 times of half cycle (from top to bottom)
0.83x 2.3x
Just enough
10x
oversampled
1.3x
Undersampled
Image courtesy of Cameron Nowell
• http://www.svi.nl/NyquistCalculator
• Pixel dimensions = Theoretical axial resolution/2.3
• Only use this if you need this resolution
Magnification NA Type 405 nm 488 nm 561 nm 633 nm
5 x 0.1 Air 704 848 975 1100
10 x 0.3 Air 235 282 325 367
20 x 0.5 Air 141 169 195 220
40 x 1.25 Oil 57 68 78 88
63 x 1.4 Oil 50 60 70 79
100 x 1.4 Oil 50 60 70 79
Nyquist-Shannon sampling
Get the best representative image
Best representation of what you are seeing on the microscope
- No offset on your raw data- Set range with your (+)/(-) control sample
- Same setting between samples- Pilot analysis before/during
imaging
Any staining can be positive or negative depending on your imaging setting
Crap in Crap out
Basic Rules of Digital Imaging
Dr Paul McMillan
Biological Optical Microscopy Platform
Raw file
Basic rules of digital data
1. SAVE
Always save as the raw file extension
.lif or .lei (Leica)
.lsm or .czi (Zeiss)
.oif or .oib (Olympus)
.dv (Deltavision)
Metadata contains all image parameters
Image metadata
• Laser wavelength• Light path settings • Objective lens / NA• Pinhole• Frame Size
• Voxel dimension• Gain/offset• Bit Depth• Scan speed• Average
Raw file
Basic rules of digital data
2. ExportImage metadata
TIF
Keeps original informationfor data analysis
JPG, GIF, PNG
Compressed, lossyfor PowerPoint only
Basic rules of digital data
3. Modification
Never process/modify on the raw files. Make a copy to process.
Raw file
Image metadata
Exported
Image
Exported
modified
Image
DataGraphs
Modification
details
Raw file
Image metadata
Exported
Image
Exported
modified
Image
DataGraphs
Modificationdetails
4. Backup
Always double back up your data (NHMRC guidelines)
Never use the Microscope computer to store your data
Image = scientific data
“oops, I was not aware of this”
“sorry, but I thought it was okay to make the images clearer”
Image beautification = data manipulation
Modification on a specific feature
Image from Mike Rossner, and Kenneth M. Yamada J Cell Biol 2004;166:11-15
Original image Manipulated image
-The gold particles have been enhanced-Background dot has been removed
Modification on a specific feature
Image from Mike Rossner, and Kenneth M. Yamada J Cell Biol 2004;166:11-15
Manipulated image ?Manipulation revealed
By contrast adjustment
Cells from other image has been added
Modification on a specific feature
➢live with it since it isn’t too bad
➢crop it out
➢resection the block and take a new photomicrograph without the tissue fold
Image from http://www.niehs.nih.gov/research/atniehs/labs/assets/docs/q_z/to_adjust_or_not_to_adjust_508.pdf
Original image Manipulated image
Modification on the background
Background Shading in one cornercleaned up
Acceptable only if- No change occurred in the actual tissue- A statement is made in the figure legend
Original image Manipulated image
Image from http://www.niehs.nih.gov/research/atniehs/labs/assets/docs/q_z/to_adjust_or_not_to_adjust_508.pdf
Modification on the background
uneven illumination & brown fat has been erased
NOT Acceptable because-large degree of background alteration (even if it is clarified and the interested area is not modified)
Original image Manipulated image
Image from http://www.niehs.nih.gov/research/atniehs/labs/assets/docs/q_z/to_adjust_or_not_to_adjust_508.pdf
Generally acceptable
Modification to whole image
Original image
Misrepresented
Modification to whole image
exposure and contrast to the maximum
Images have not been processed identically
Completelyblack
colour balance differs
(-)Control (+)control sample
black
Image from Mike Blatt, and Cathie Martin Plant Physiol. 2013;163:3-4
GFP expression on the sample looks similar to (+)control
Image presentation
Dr Paul McMillan
Biological Optical Microscopy Platform
Data presentation guidelines
Cell – Instructions to authors
• Any alterations should be applied to the entire image.
• Clearly explain all alterations in the figure legend.
• Only compare data that are appropriate to compare (e.g., data from the same experiment, acquired the same way).
• Individual images should not be used in multiple figures (unless the figures describe different aspects of the same experiment)
My 5 cents worth
• Be aware of colour sensitivity & colour blindness
• Be aware of printing issues
Colour
• (Most) Fluorescence detectors are monochrome
• Images are usually grayscale
• Converted to colour using Look Up Tables (LUTs)
Blue
Green
Red
0 255
Colour sensitivity
RGB RED GREEN BLUE
0
255
0
255 0
00
2550
0
0
255 255
255255
RGB vs CMYK
http://wip.blackfox1985.com/wp-content/uploads/2016/03/rgb-vs-cmyk.jpg
Colours – RGB vs CMYK
http://www.colorprintingforum.com/attachments/layout-graphics-prepress-software/406d1254902900-how-change-rgb-cmyk-cmyk-vs-rgb.jpg
Use grayscale for panels