Finalised

updated links and formating; included citations for papers mentioned

Author: BrentonCavanagh

_pages/imaging/colocalization-acquisition.md

@@ -1,5 +1,4 @@
 ---
-mediawiki: Colocalization_-_hardware_setup_and_image_acquisition
 title: Colocalization - hardware setup and image acquisition
 categories: [Colocalization,Tutorials,Color processing]
 section: Imaging
@@ -50,9 +49,9 @@ There are various problems that need to be understood and overcome while collect
 
 ![](/media/imaging/colocblur.gif)
 
-*`Definition:`*` The Fourier spectrum of the image lacks proper high frequencies. `
+*`Definition:` The Fourier spectrum of the image lacks proper high frequencies.*
 
-*`Definition 2:`*` The small features/edges of the real object are not well represented in the image `
+*`Definition:` The small features/edges of the real object are not well represented in the image.*
 
 This phenomenon will artificially enlarge the size of your objects, making them look bigger. The problem with this is that two non colocalizing objects might appear to colocalize in the image because of the blur. This gives false-positive results.
 
@@ -73,15 +72,15 @@ Some things we can't do anything about (hey, that's the *light* microscopy facil
 
 -   Diffraction: The spatial resolution of the conventional widefield or confocal (as typically used - noise limited) light microscope is limited by diffraction, and is dependent on the wavelength of the light (shorter is better) and the numerical aperture (NA) of the objective lens (higher is better). XY resolution is better than resolution in Z direction (about 3x better for a high NA objective. Spatial sampling should be done so that there are not less than and not too many more than 2.3 pixels across the resolution limit (according to Nyqvist). For a high NA lens (say a 63x 1.4 oil immersion) this means the pixels in XY should be about 80-100 nm, and in Z about 250 nm. Also see Sampling below.
 
-If the blurring of you images is too much, you might want to consider [deconvolution](/imaging/deconvolution). Have a look at [this page](http://support.svi.nl/wiki/BlurAndNoiseAffectColocalization) from SVI (the guys that provides [Huygens](Huygens) software to see how it can help.
+If the blurring of you images is too much, you might want to consider [deconvolution](/imaging/deconvolution). Have a look at [this page](http://support.svi.nl/wiki/BlurAndNoiseAffectColocalization) from SVI, the guys that provides [Huygens](https://svi.nl/) software to see how it can help.
 
 
 
 ### Background
 
 ![](/media/imaging/colocbgnoise.gif)
 
-*`Definition:`*` Unspecific signal. `
+*`Definition:` Unspecific signal.*
 
 The presence of a high background will generate a huge amount of artificial, irrelevant colocalization. You are not interested in unspecific colocalization - your reviewers might only be mildly happy if you spend a lot of time analysing background. You do not want background to affect your analysis results.
 
@@ -100,7 +99,7 @@ Getting rid of high background? Consider:
 
 ### Noise
 
-*`Definition:`*` Uncertainty of signal. `
+*`Definition:` Uncertainty of signal.*
 
 Colocalisation in a pixel might be false, coming from noise, and thus, not be real.  For correlation based statistical approaches to colocalisation analysis, increasing noise lowers the apparent colocalisation. Hence, for noisy images, an attempt must be made to suppress or remove the noise in order to get closer to the true noise free correlation. Constrained iterative deconvolution can act as a smart noise filter that increases contrast and suppresses noise, so widefield and confocal image data should be deconvolved before colocalisation analysis.
 
@@ -117,9 +116,9 @@ One must have a signal strong enough to allow a good enough signal:noise ratio,
 
 ### Cross talk and Bleed through
 
-*`Definition 1:`*` Fluorophores do not match optical components (excitation / emission filters, lasers, dichroics). `
+*`Definition 1:` Fluorophores do not match optical components (excitation / emission filters, lasers, dichroics).*
 
-*`Definition 2:`*` You detect emission from the wrong dye, and falsely believe it comes from the right dye.`
+*`Definition 2:` You detect emission from the wrong dye, and falsely believe it comes from the right dye.*
 
 So... this one is *the* very worst and most dangerous problem in colocalization experiments. It can be explained with the following spectra:
 
@@ -130,16 +129,16 @@ In this picture, the detection setup is configured so as to measure emitted ligh
 
 You could avoid this by selectively only exciting fluorophore number one, and acquire the first image, then excite the fluorophore number 2 and acquire the second image (sequential imaging). But on these spectra, you can see that the excitation curves also overlap, and that the excitation wavelength hits both dyes efficiently. The two fluorophores will both be excited at this wavelength; a process called **cross-talk**. Your last hope is that there is a laser line that hits the first excitation spectrum without touching the other, and reciprocally for the second one.
 
-Dye choice is critical here. Avoid DAPI and GFP, since DAPI emission is very broad, and bleeds through into GFP detection. Better to choose a far red nulclear/DMA stain like Draq-5 for use with GFP. Choose dyes by looking at their excitation and emission spectra, and pick dyes that match your microscope fluorescence filters and laser excitation lines (eg. look at the [Invitrogen Molecular Probles Java Spectra Browser](http://www.invitrogen.com/site/us/en/home/support/Research-Tools/Fluorescence-SpectraViewer.html) for doing that). Choose dydters that are well spectrally separated it at all possible. Different microscopes in the LMF have different choices of excitation filter and laser line wavelengths, and different possibilities for emission filters. Choose your dyes carefully. 
+Dye choice is critical here. Avoid DAPI and GFP, since DAPI emission is very broad, and bleeds through into GFP detection. Better to choose a far red nulclear/DMA stain like Draq-5 for use with GFP. Choose dyes by looking at their excitation and emission spectra, and pick dyes that match your microscope fluorescence filters and laser excitation lines (eg. look at the [Thermo Fisher Scientific Fluorescence SpectraViewer](https://www.thermofisher.com/order/fluorescence-spectraviewer) for doing that). Choose dydters that are well spectrally separated it at all possible. Different microscopes in the LMF have different choices of excitation filter and laser line wavelengths, and different possibilities for emission filters. Choose your dyes carefully. 
 
 
 
 
 ### Poor Spatial Sampling
 
-*`Definition:`* `The pixel size does not allow for a highly spatially resolved colocalisation analysis.`
+*`Definition:` The pixel size does not allow for a highly spatially resolved colocalisation analysis.*
 
-*`Definition 2:`* `Pixels are too big. Close but separate objects which do not colocalise, appear to colocalise.`
+*`Definition 2:` Pixels are too big. Close but separate objects which do not colocalise, appear to colocalise.*
 
 As a general rule, you should have each object you image sampled over many pixels. When this is not the case, a colocalization statement cannot be reliable made, as the assumptions of the method no longer hold. 
 
@@ -152,26 +151,30 @@ Also remember: since we are doing light microscopy, two objects closer and small
 
 ### Detector Saturation
 
-The problem with "saturated pixels" occurs mainly when we use pixel-based colocalization (*i.e.* when we are interested in intensity correlation over space). It is possible to derive the quantitative error made with saturation. See [here](Colocalization_ErrorDueToSaturation). If pixels/images are saturated (that is having pixels with an intensity level at the top of the range, ie 255 for 8 bit data) then they are missing information about the real spatial intensity distribution in the sample. This is the most important data in your analysis, as you are usually most interested in the brightest objects - right?! That means its a really bad idea to throw that information away when you collect the images.
+The problem with "saturated pixels" occurs mainly when we use pixel-based colocalization (*i.e.* when we are interested in intensity correlation over space). It is possible to derive the quantitative error made with saturation. See [here](colocalization-analysis#check-your-image-quality). If pixels/images are saturated (that is having pixels with an intensity level at the top of the range, ie 255 for 8 bit data) then they are missing information about the real spatial intensity distribution in the sample. This is the most important data in your analysis, as you are usually most interested in the brightest objects - right?! That means its a really bad idea to throw that information away when you collect the images.
 
 ### Chromatic Shift
 
 If the microscope has poor quality or non chromatically corrected objective lenses and / or has other alignment problems, then images of the same object in different colour channels will appear in different places. Even expensive lenses have a little residual chromatic shift. This is very bad, as it means that, especially for smaller objects, you miss some of the real colocalisation. The microscope should be checked with multi colour 1 or 0.5 micron bead samples before imaging to see if there is a significant problem. Images can be corrected for x y and z chromatic shift if a bead image has been taken under identical conditions.
 
-For more details and possible chromatic shift measurement and correction strategies see [Chromatic\_shift\_origins\_measurement\_and\_correction](/imaging/chromatic-shift)
+For more details and possible chromatic shift measurement and correction strategies see [Chromatic shift origins, measurement and correction](/imaging/chromatic-shift)
 
 ### Conclusion
 
 **Think about your biological experiment, dyes and imaging, and be careful throughout. Collect image data that is suitable for quantification.** For example, Tetraspeck fluorescent beads with 3-4 colors and different sizes (small enough to check shifts) make great positive controls.
 
-Also, for a review of the hardware part, read *Zucker, 2006, Cytometry*.
+Also, for a review of the hardware part, read *Zucker, 2006* [^1]. 
 
-**`Credits`**  
-`From Jan Peychl notes. `  
-`The two animated gifs are from the Huygens Software wiki page. `
+
+>**Credits**  
+>From Jan Peychl notes.   
+>The two animated gifs are from the Huygens Software wiki page. 
 
 # Further reading
 
-[The 39 steps: a cautionary tale of quantitative 3-D fluorescence microscopy](http://www.ncbi.nlm.nih.gov/pubmed/10818693) by J. Pawley.
+*The 39 steps: a cautionary tale of quantitative 3-D fluorescence microscopy* by J. Pawley [^2].
+
+{% include citation fn=1 doi='10.1002/cyto.a.20314' %}
+{% include citation fn=2 doi='10.2144/00285bt01' %}