Microphotography: How'd they do that?

Here is a link to an article in the current on-line version of Wired. It features the winning entries of this year’s Nikon’s Small World microscope photography competition. Many of the images are just spectacular.

I do have a question, though, about this image of fruit fly larvae. On the main page, it’s captioned as having used a ‘confocal’ technique at 100X. Is that technique responsible for the exquisite color detail we see in the “fingers” and “head parts”? Or, were the color scheme (and results) introduced into the image by means of image processing software (or even, perhaps, simply as the product of the imagination of the person who ‘developed’ the pic)?

Bottom line questions are*: are we seeing the results of optics or software?; if we shrunk ourselves down to that scale, would what we see look like what we see in the picture?

*We could probably ask this about all of the entries

You can get fluorescing dyes and stains that bind to specific proteins, and you can use them to label parts of cells or organisms, which will then fluoresce under UV illumination.

So you could wash your larvae with the dyes/stains and when you rinse the stain off it will only stay in the region of the larvae which is expressing a particular protein.

A lot might have changed in the 15 or so years since I did any serious microscopy, but that is what I think has been done in that image.

Well, to the naked eye, larval fruit flies are white…

Confocal imaging does not produce a ‘true’ colour image, it’s a technique I don’t really understand using fluorescence, but some of the other pictures appear to be simply close-ups, using something reasonably close in spectrum to natural lighting.

Thanks for the link though, there’s some fascinating stuff there…

Well, if this isn’t one of those moments of Straight Dope serendipity. I do fluorescence microscopy on fruit fly larvae, and I’m taking a graduate-level class on microscopy right now. So I’ll go ahead and claim to be qualified to answer this.

So, what you’re looking at is almost certainly a false color image. In fact, let’s just say certainly. But that doesn’t mean it’s just pretend. If you looked with your eyes through the microscope, and all of the illumination lasers were firing, well, you’d go blind. But if you didn’t, or if you used a less intense illumination, you would see more or less what you’re looking at in the photo, with a few caveats. First is that the colors may be completely different. The balance between the colors could be different, too.

The confocal part is just a method of eliminating out-of-focus light, which gives a very bright, crisp, clean picture with a very good signal to noise ratio. It doesn’t really have anything to do with the color, but it does help with the crisp resolution of fine details.

As far as colors go, the larva is stained with it looks like four separate dyes, conjugated either to a protein directly or to an antibody against some specific protein. You can see blue, green, orange, and red signals. The blue signal looks like it’s some mouth hook protein; I think the green is lighting up some neurons, perhaps; etc. When you take the image, each dye is excited and recorded one at a time. There’s a blue channel, green channel, etc. Now, again, those dyes really DO have real colors. However, the signal that’s recorded by the camera (actually a series of photomultiplier tubes, but it’s the same principle), and sent to the computer is just a number for each pixel, which gives a black and white image. For each pixel, you’re simply detecting an intensity value ranging from 0 (black) to 255 (white) (for an 8-bit camera), with shades of grey in between. This value is directly related to the number of photons emitted by that pixel as it was being scanned.

So you do your scan with four different channels. For each pixel, you get four different greyscale black and white values, which the computer displays as a black and white image. You can manipulate these in all sorts of ways, of course, to enhance contrast or bring out detail. This can be done differently to each channel, which is why I said that we may be seeing a different balance of colors than was “really” there. They may have, for instance, boosted the green signal relative to the red.

And then you apply a false color to each channel. If you pick blue, for instance, 0 still corresponds to black, but 255 now corresponds to some really intense blue color that you can select. The 254 levels in between will be represented by a range of dark blue to bright blue. The thing is, the color you select to apply to the image doesn’t have to be the same color as the original dye in the larva. Often times it’s not even remotely similar. When preparing images for publication, the goal is to make your information clearly visible to the reader. Far red and far blue are difficult for the human eye to pick out, so they’re often not used, even if the original dye was red or blue. There’s also more of an effort these days to make your colors colorblind-friendly. Magenta and yellow-green is a common combination. In this case, their goal was to make a pretty picture that make people go “Ooooh!”, so that no doubt influenced their choice of colors.

Anyway, yes, the larva really did look like that, as long as it had light of the right wavelengths illuminating it, and through the right filters. More or less. We’re definitely not seeing anything that was introduced by the imagination of the researchers.

Very helpful explanations - thanks, all!

So, basically, the structure and anatomy shown in the picture are faithful, accurate depictions of reality. On the other hand, the colors used to delineate the structure are arbitrary and don’t correspond to anything in particular, i.e. the colors were chosen simply on their basis to look cool and/or to highlight the structures of interest.

It reminds me of the coloring schemes and processes used to depict Mandelbroit and Julia sets (and related entities), e.g. 1 & e.g. 2

That’s Whole Mount Confocal microscopy. Similar to regular confocal but you don’t need to section the specimen.

We use it all the time to image insects like Drosophila and mosquitoes. It’s a lovely technique (gotten us a few journal cover images).