NetNotes
on the eye icon (in the Layers palette) to turn off and on the layer, one can see the subtle effect of sharpening and consequent improvement in local contrast. Tis effect works much like selenium toning on photographs to improve acutance. Sharpening should be done before setting tonal levels, and to a duplicated image, not the original or raw image. Finally, in my own experience, the blackest tones on photographic paper exceeds that of inkjets, and the darkness of the black tones has a marked effect on the perception of contrast. Tus the necessity for more than one shade of black ink in an inkjet printer, as John had mentioned, and the necessity for sending out to an agency as mentioned in an earlier post. At one time the U of Minnesota core facility owned a Fuji Pictrography 3000, which is a dye-sublimation/ photographic processing device. Tis could achieve nearly the black tones provided on Kodak RC paper, and side-by-side comparisons with photographic paper could, in some instances, be perceived as “better” than the wet-processed papers, depending on the sample. A means for providing a greater perception of contrast can also be done by blue-shiſting the image. As microscopists, we develop a “color memory.” Even though photographic paper is “black and white,” the Agfa papers tended to be bluer than the Kodak papers (which looked more brown under artificial light), and we became not only accustomed to seeing that color shiſt, but became judges of quality based upon the color. Try it sometime: make the image into RGB Color, and then use the gamma slider in Levels to adjust the blue channel, and then print. Jerry Sedgewick
sedge001@umn.edu Tue May 18 I have been enjoying the discussion about digital and film, but
I want to make comment here: Digital image acquisition is no silver bullet. You cannot recover digitally what you don’t have physically. To have a good digital image, one needs to start out with a good physical signal. While digital imaging can help with difficult situations and make a small contrast more visible, one has to have a contrast to begin with. What I want to say is that digital imaging does not absolve the TEM user from knowing the instrument and its limitations and capabilities. Mike Bode
mike.bode@
resaltatech.com Tue May 25
Image Processing: deconvolution & confocal microscopy I am a first-year graduate student trying to get started with
co-localization analysis of IF images taken by confocal microscopy. I am working with fixed cells and mainly trying to measure co-localization of 2 proteins. My main questions are concerning the use of z-stacks and deconvolution: 1. Are z-stacks always necessary for measurement of co-localization, or are single images sufficient? 2. I understand that deconvolution is essential for wide-field images, but does it improve results that much when you’re already using confocal? 3. If you would recommend deconvolution, is the ImageJ plug-in a good one to use? Tank you for your time & help. I have read many articles on this topic, but I keep ending up with conflicting answers. Molly Shaw mshaw1@
lumc.edu Scientists have published co-localization results using single
planes from confocal microscopes, z-series, z-series with deconvo- lution and results from widefield microscopes with thinly sliced specimens. Tis alone creates confusion about the “right” method. Do yourself a favor and go through the effort of deconvolving images. You will then have a greater likelihood of being published in a high profile journal, although you might be asked for more robust methods like Fluorescence Lifetime Imaging Microscopy (FLIM) or Forster Resonance Energy Transfer (FRET). Confocal microscopes do collect out of focus fluorescence, and there is elongation in the z-axis, and deconvolving images ameliorates those problems. If nothing else, do the following: Use a high N.A. lens (1.4 or better). Mount the specimen in a medium that has a refractive index of 1.4 or better. Use a coverslip
66
thickness intended for the oil objective (0.17 mm). Make sure you don’t just place one image over another (red + green = yellow—presto! Co-localization!); show results using scatterplots. If the feature you are labeling is at near subresolution dimensions, then use Nyquist rates for collecting (your core microscopy lab can provide info about Nyquist rates for their equipment, or at least a way to determine these). I have tried Volume J a few years back and it crashed with large image sets. I suspect computers/soſtware have caught up by now. Certainly Image J/Volume J is good for attempting deconvolution. It’s advisable to make a point spread function (psf) from subresolution beads (0.8 microns, or what you read in the literature as the bead to use for the objective: I can’t remember ofand what I used way back when) and then deconvolve using the psf. Have fun! Jerry Sedgewick
sedge001@umn.edu Tue Jun 8 You are right that a confocal will not benefit from deconvolution
soſtware anywhere near as much as a widefield microscope, for the reason you suggest, i.e. much of the out-of focus information is removed by the confocal iris/pinhole in any case. I use Improvison’s [Perkin Elmer] 3D Volocity soſtware to look at things like 3D images, 3D cellular volumes and cell morphometry. In order to get the best from a Z slice you are advised to over sample the z-stack, i.e. for a 0.9 µm optical slice capture the z stack at least 0.45 µm apart. Our Zeiss 510 has a button to automatically ‘double oversample’ like this. For the likes of Volocity deconvolution, Improvision recommended to me that you massively oversample the confocal z-stack with ten or more slices to every 0.9 µm z focus step. I think the rep suggested perhaps up to as high as 100× oversampling. I rarely use the optional ‘Restoration’ module [deconvolution] Volocity module as I only use the confocal for z-stack, and double oversampling and 3D reconstruction alone with Volocity is adequate for my needs. For co-localization I would initially try and tackle the co-localization quantification in 2D using MetaMorph probably just because it’s just easier. I don’t use ImageJ very much as I have a MetaMorph license, and we don’t have the 3D Quantitation module for Volocity in any case, that might be needed for 3D co-localization quantification—if interested just ask Improvision about that as their support is superb. Always be suspicious that any co-localization isn’t bleedthrough. You can check this via spectral un-mixing on a 510 Metahead or just try imaging single labeled samples of each fluorochrome and check there’s no signal in the other channels where there is now no longer any label at the same confocal imaging settings. Don’t’ forget a sample with no label just in case of autofluorescence. To quote Improvisions own website: “If you acquire your images from a widefield microscope and want to volume render or make measurements from the image data, you will need to use an image restoration technique [de-convolution] to remove the out-of-focus information—a product of the optical properties of the microscope. Even if your images are captured using a laser scanning or spinning disk confocal microscope, the image quality may benefit from image restoration.” Te word ‘may’ is probably significant. Keith J. Morris
kjmorris@well.ox.ac.uk Tu Jun 10 Be very careful with quantification in fluorescence. A higher
signal doesn’t necessarily mean a higher concentration of the antigen, it may also mean a better antibody, a better accessibility for the antibody, a better stability from the laser light and so many other things. You could trace a line through the structure and show a profile scan (confocals are good at that), this is a nice way to concentrate the attention only on one structure and demonstrate the co-localization of 2 peaks of intensity. Te profile also allows a fine analysis of the both signals intensities and who knows, perhaps you’ll discover that both signals are not just perfectly superimposed, but are next to each other (very close), or that one signal has more a ring pattern and so on. Small differences in intensities are not easily
www.microscopy-today.com • 2010 September
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42 |
Page 43 |
Page 44 |
Page 45 |
Page 46 |
Page 47 |
Page 48 |
Page 49 |
Page 50 |
Page 51 |
Page 52 |
Page 53 |
Page 54 |
Page 55 |
Page 56 |
Page 57 |
Page 58 |
Page 59 |
Page 60 |
Page 61 |
Page 62 |
Page 63 |
Page 64 |
Page 65 |
Page 66 |
Page 67 |
Page 68 |
Page 69 |
Page 70 |
Page 71 |
Page 72 |
Page 73 |
Page 74 |
Page 75 |
Page 76