HiLo Microscopy

Comments
Add One

There’s a stark contrast between the elegant simplicity of a conventional widefield microscope and the much more complex apparatus need for point-by-point scanning in confocal microscopy. It’s this point-by-point scanning, together with a pinhole, which gives the confocal microscope its optical sectioning ability. By removing the out of focus blur which would degrade a conventional microscope image, confocal microscopes obtain crisp, clean images of thick samples, or even of in vivo tissue. The additional complexity involved with confocal operation, which includes the requirement to use a laser rather than a thermal light source, is accepted as the price that has to be paid if we want to obtain these kinds of images. But now, HiLo microscopy is offering an alternative approach which could have a number of niche applications, particularly where space or cost preclude the use of a full confocal microscopy setup.

There are ways of obtaining depth sectioned microscopy images which don’t involve point-by-point scanning or even necessarily require laser illumination.  There are several variations on the theme of ‘structured illumination’ which, in this context, means projecting a pattern of illumination (usually a grid-like pattern) onto the sample. The pattern is projected with the same lens that’s used for imaging, resulting in an illumination pattern which is in focus at the imaging focal plane, and blurred everywhere else. This provides, at least in some abstract sense, a way of distinguishing between the in-focus and out-of-focus components of the image. The in-focus parts of the image are illuminated with a high fidelity pattern, while the out-of-focus parts are illuminated by a blurred pattern. For layers of the sample which are sufficiently far from the focal plane, this blurring will be so severe that the illumination is essentially uniform.

The devil, as is often the case, turns out to be in the detail. In order to extract an artefact free sectioned image we have to acquire a sequence of three structured images, each with a precise (and known) shift of the illumination pattern. By a process of subtracting images from each other it’s possible to cancel out all the out-of-focus image components – leaving only the in-focus optical section. However, there are noise problems associated with performing subtractions between images, and artefacts begin to creep in if the sample is moving. This makes reliable structured illumination imaging more of an engineering challenge than it may first appear. Whether this can fully explain the reluctance of the market to embrace structured illumination microscopy as a real alternative to confocal microscopy is less certain.

A Speckled Approach

In the last few years Jerome Mertz’s Group at the University of Boston has been working on a slightly different idea. Their approach is to project a random speckle pattern onto the sample using a laser and a diffuser. In a similar way to the conventional grid-type illumination, the speckle pattern appears in-focus and with high fidelity at the focal plane, while everywhere else it becomes blurred. In fact if we go far enough away, we won’t see a pattern any more, and the illumination is uniform.

So now when we image the sample we get an in-focus but speckled image of the plane of interest, super-imposed on blurred but speckle-free images of all the other layers. At first sight it seems that we have not achieved very much at all  – in fact we have actually degraded the very part of the image we want to retain! But we have actually achieved something very important; we have introduced a distinction between in-focus and out-of-focus components – one is speckled and the other is not. Now we need to find a way to exploit this difference so that we can discard the out-of-focus parts.

The most obvious way of doing this, and the first published by Mertz1, is to acquire lots of different images, each with a different speckle pattern projected onto the sample. It’s then a simple matter to make a measurement of how the intensity of each pixel changes with time, by measuring, for example, the standard deviation of its intensity. Signal from out-of-focus planes won’t contribute to this variation, since the points are uniformly illuminated regardless of the speckle pattern. So the variation is now a measure of the in-focus signal at that pixel.

The problem with this approach is that it requires the acquisition of a large number of raw speckle images just to obtain just one sectioned image. In fact the authors found that they needed about 30 images to get a high quality optical section. Even with a good scientific camera that is going to take a significant fraction of a second.  Given that pretty much any sample motion during the acquisition is going to ruin the measurement, the method begins to looks pretty unattractive for any kind of in-vivo work.

In Space

So if measuring the speckle variation temporally isn’t going to work, can we measure it spatially instead? The answer is ‘yes we can’, but with a caveat that we degrade the spatial resolution by removing higher spatial frequencies – rarely a good thing in microscopy. What we need is some way to recover the high spatial frequency information from the in-focus plane.

It turns out that it can be done pretty easily, thanks to the fact that all of the high spatial frequencies in a normal widefield image originate from the in-focus plane. High frequency spatial variations from out-of-focus planes have already been removed by the imaging system – that is why the images of these planes look blurred. So if we apply a high pass-filter to a conventional widefield image (i.e. one without any speckle pattern), we get an image in which all the out of focus contributions have been removed

But hang on, you might say, doesn’t that mean we now have obtained a sectioned image just by simple filtering? Don’t throw away your confocal microscope just yet. While it’s true that we now have a sectioned image, it’s one in which all the low frequencies from the in-focus plane have also been removed. In general this isn’t particularly useful, and it certainly isn’t going to look very pretty.

But now we have gone full circle; using spatial speckle averaging we can create a low pass sectioned image, and using simple spatial filtering of a widefield image we can a high pass sectioned image. So all we have to do is put those images together and we have our sectioned image with all the spatial frequencies present.

Of course in practice the process of actually combining these image is a little more complex, but Mertz et al. describe the procedure fully in their papers2,3. An ImageJ plug-in which performs the combination, together with a couple of raw images, is now available on the University of Boston website. Figure 1 shows the one of the example images included. Sadly they’ve only released the compiled classes, so if you want to play around with the code you’ll either have to ask nicely or recreate it based on what’s in their papers. You also have to work out how you are going to switch between speckle and conventional illumination, although that turns out not to be too difficult to implement.

Example of HiLo depth sectioning, showing a widefield and a sectioned image.

Figure 1 : Example images from the HiLo ImageJ Plug-in, showing a widefield image (left) and reconstructed sectioned image (right).

The Boston group have also shown that you don’t have to use a speckle pattern, you can use a grid similar to those used for structured illumination microscopy (both the depth-sectioning and super-resolution kinds). This makes it possible to use the HiLo technique endoscopically, with optical fibre bundles4. No in-vivo results have appeared yet, so it remains to be seen if motion artefacts will degrade the images, although the Group suggest that a double-exposure camera could help matters.

Coming to a store near you?

A company, Hilo Microscopy Inc. has been set up to exploit the technology. It looks as though their tactics will be to market it as a cheap (~$10 000) bolt-on to existing microscopes, putting it in direct competition with Aurox, a spin-out from Tony Wilson’s Group at Oxford. The Aurox is single-shot and requires only very simple processing, so it’s not immediately obvious what advantages the HiLo product will have, other than the possibility of tuning the depth sectioning. It will be interesting to see if a device appears on the market sometime soon.

References

  1. Ventalon, C. and J. Mertz, Quasi-confocal fluorescence sectioning with dynamic speckle illumination. Optics letters, 2005. 30: p. 3350-2.
  2. Lim, D., K.K. Chu, and J. Mertz, Wide-field fluorescence sectioning with hybrid speckle and uniform-illumination microscopy. Optics letters, 2008. 33: p. 1819-21.
  3. Lim, D., et al., Optically sectioned in vivo imaging with speckle illumination HiLo microscopy. Journal of biomedical optics, 2011. 16: p. 016014.
  4. Ford, T.N., D. Lim, and J. Mertz, Fast optically sectioned fluorescence HiLo endomicroscopy. Journal of biomedical optics, 2012. 17: p. 021105

Leave a Comment

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>