Microscopy through Fibre Bundles

Add One

Microscopes are great tools – they make modern biology possible, and are essential for the diagnosis of many diseases. But they have one big disadvantage when it comes to medicine: they can’t be used directly on patients. Or at least they couldn’t, because now a new technology has emerged that allows us to perform ‘endomicroscopy’ – microscopy inside of the patient. This in vivo microscopy has lots of potential applications and is a pretty exciting area to be working in. The technology still has a few problems, but it’s shaping up to be one of the key new imaging techniques of the twenty-first century. And it has become practical because of the use of optical fibre bundles.

Depth of Focus

At first sight, endomicroscopy may not seem all that impressive. After all, we’ve been performing endoscopy (that is, macroscopic imaging) inside a patient for decades. Surely in vivo microscopy is just a case of adding a microscope objective lens, perhaps some filters for fluorescence imaging, and we’re away!

Unfortunately, it’s not quite the sample. The root of the problem is the small depth of focus that microscope objectives give us; meaning that when a relatively thick sample is placed under a microscope, only one particular depth layer will be in focus at any one time. The phenomenon of depth of focus (or depth of field) is familiar to anyone who has used an expensive camera, but in microscopy the effect is much more severe. In fact the in-focus layer tends to be only a few microns thick; meaning that for translucent samples there are likely to be a lot more out-of-focus parts than in-focus parts.

While only this thin layer is in-focus, light returning from other layers will still be collected by the microscope and projected onto the camera. This means that image will tend to have a large background blur (from the out-of-focus layers) super-imposed on the image of the in-focus layer that we are  interested in. This problem can be so bad that the images become useless, especially if we are looking underneath the top surface (which is almost always the most reflective layer).

The problem can be solved in one of two ways. The first is to slice the sample very thinly, so that the whole of the sample can be in focus at the same time. This works pretty well, and is the standard technique used in histology to study tissue samples excised from the patient. But obviously it can’t used in vivo, directly on the patient, since they tend not to come ready-sliced! It requires an invasive biopsy which has its own disadvantages, particularly for certain sensitive organs.

The other solution is the confocal microscope. A quick google search will show lots of pages explaining how it works, and I’d also recommend Tony Wilson’s book if you want a deeper discussion. The general idea is that a confocal microscope produces an ‘optical section’ – it generates an image of a single depth layer without having to physically slice the tissue. It works by imaging the tissue point by point, and imaging the collected light through a pinhole. Any light which is out-of-focus won’t be focused correctly onto the pinhole, and so will be rejected. Light from in-focus layers of the tissue passes through the pinhole and reaches the photodetector.

So at first glance the confocal microscope it is an ideal candidate for in vivo use, as we no longer have to worry about the out-of-focus blur. Unfortunately, it’s considerably more complex, and so usually more bulky, than a conventional microscope. One complication is that the image is built up by scanning a single illumination point across the sample in a raster pattern, rather than capturing the whole image at once. In table-top devices this is easy to do – mechanical scanning mirrors can rapidly scan the illumination point to generates high frame rates. But the need to scan is a huge impediment to making a microscope small enough to fit inside a patient.

For completeness, I should point out that it is possible to miniaturise the scanning mirrors using MEMS technology, and that this has been done for endomicroscopy1 as well as for optical coherence tomography (OCT). The idea is to transmit light in and out of the patient along an optical fibre, to a small probe-head containing the scanners. The technology actually works reasonably well, although the probes tend to be relatively large and expensive, and it requires high voltages to be passed along wires inside the patient.  These problems can be overcome using fibre bundles.

Into the Fibre

Fibre bundles are interesting devices.  A normal optical fibre consists of three concentric layers. The inner layer is known as the core – it is through this that most of the light actually travels. The next layer is called the cladding. This has a different refractive index to the core, which results in the light being constrained within the fibre and travelling along it, rather than escaping in all directions. Finally, there is the cladding, which protects the otherwise delicate fibre.

Fibre imaging bundles are not simply lots of optical fibres stuck together. They consist of a single, shared cladding, and a large number (often tens of thousands) of individual cores within the cladding. Each core acts as an individual light conduit, confining light to travel along it, just as in a normal optical fibre. We say that the fibre bundle is ‘coherent’ if the cores do not cross over at any point – so that the position of each core at one end of the bundle is the same as at the other. If the bundle is coherent then it can be used to transfer an image from one place to another, along a flexible path, with each core of the fibre effectively acting as an image pixel.

The idea of using fibre bundles for microscopy goes back to 1993, and a paper by Gmitro and Aziz2. Their idea was simple. They set up their fibre bundle so that one end lay on the image plane of a table top confocal microscope. The other end of the bundle was then used to image tissue via a microscope objective. The raster scanning needed for confocal imaging was performed outside of the patient, and transferred to the imaging site using the bundle. The bundle then transferred the image back to the microscope.

Since 1993 a number of improved versions of this device have been implemented. Fibre bundle endomicroscopy has been demonstrated as a viable imaging technique, and has even been commercialised. But fibre bundles are certainly not a panacea when it comes to in vivo imaging.

The biggest disadvantage of using fibre bundle is that the resolution of the image is limited by the spacing of the fibre cores. To some extent we can overcome this by using magnifying optics at the distal (i.e. patient) end of the bundle. This essentially means that the core spacing is ‘minified’ to a smaller size on the tissue. But this also reduces the field of view by the same factor, so we can only go so far with this before we end up with images covering too small an area to be useful. There are also physical limits to how high a magnification lens we can pack into such a small probe without running into other problems, particularly optical aberrations. So practical fibre bundle endomicroscopes are unlikely to offer resolution better than a couple of microns.

The other main problem with using fibre bundles is that reflectance (as opposed to fluorescence) microscopy is a little more difficult due to the strong reflections from both ends of the fibre bundle. Unlike most of the other optical elements in the microscope, both of the fibre bundle ends lie on what are known as conjugate image planes. These are planes which are imaged onto the confocal pinhole, meaning that any reflections survive to pollute the image. Since fibre-air interfaces are generally more reflective than tissue samples this means that the image is overwhelmed by a strong background.

A practical solution to this problem was suggested by Juškaitis et al., from Oxford University3. Firstly, they angle-lapped both ends of the fibre bundle so that most of the reflections would not be coupled into the microscope. They then used a cover slip and index matching gel at either end of the bundle to shift the reflective interface away from the conjugate plane. Although the reflections still occur, they are now effectively ‘out of focus’ and are rejected by the confocal pinhole. Using this approach the Oxford Group were able to obtain good quality reflection mode images. But they also found that they achieved better results when using an incoherent light source and a Nipkow spinning disk instead of a laser scanning unit.

Despite their success, most implementations of fibre bundle microscopes have operated in fluorescence mode. This is probably due to the better contrast obtained with fluorescent staining rather than any practical problems in implementing Juškaitis et al.’s method. The possibility for a dual mode endomicroscope, which can obtain both fluorescence and reflectance images, remains seemingly unexplored.

By 2006, fluorescence fibre bundle endomicroscopy had been commercialised in the form of Mauna Kea’s Cellvizio system. This currently offers excitation wavelength of 488 nm (the clinical system) and 660 nm (the pre-clinical system). It also operates with a range of probes which have different magnification factors at the distal (i.e. patient) end. This allows the user to select the resolution/field of view/imaging-depth trade off which is most suitable for their application. Their device has been trialled for a range of applications, but is used mainly in the gastro-intestinal tract. The Cellvizio website outlines most of the applications and has links to some relevant literature.

Away from Confocal

It’s not only point-scanning confocal microscopy which we can use fibre bundles for. Widefield (i.e. conventional, non-depth selecting) fluorescence microscopy is also a possibility, and indeed there is video tutorial at JOVE on how to make your own! This is less suitable for in vivo imaging though, as the out-of-focus blur will tend to make images look very unclear, and so it only really has a few niche applications.

A group working in Arizona have spent several years developing a line scanning endomicroscope in which the image is formed directly on a scientific grade CCD camera4,5. This provides reduced depth sectioning compared with point scanning microscopy, but does simplify the optical system and allow for potentially very high frame rates. They can also configure their device to capture spectroscopic data – collecting complete wavelength spectra for a line of points on the sample6. This seems to work quite well, although it’s questionable if there is much practical benefit in having such complete spectra when all we are really measuring is the (known) response of fluorophores to excitation.

A few groups have also tried to perform OCT using fibre bundles7,8. A full discussion of this is probably for another post, but suffice it to say that the results have not been particularly encouraging. One problem is that OCT is very sensitive to dispersion and polarisation, something which is very difficult to control when using fibre bundles.

Problems Ahead

Although it’s come a long way, endomicroscopy isn’t yet a fully mature technology. It’s been under development for almost twenty years now, but there is still some work to be done before it becomes a fully mainstream tool.

One of the biggest problems is the need to add exogenous stains to the tissue. If we excise tissue then there is no problem in applying whatever stains or dyes we want in order to get good contrast in our images. But that’s not a luxury we have for in vivo imaging. Only a handful of fluorescent stains are approved for use in patients, and although this number may increase, there is clearly never going to be a large a range as we might wish.

There are also problems with penetration depth. Slicing ex vivo tissue doesn’t only allow us to remove out-of-focus blur, if also lets us look deeper into the tissue. Endomicroscopy gets us, at most, a millimetre into the tissue, and usually substantially less. Techniques such as OCT and two-photon microscopy may increase the penetration a little, but realistically no optical microscopy technique is going to allow us to probe more than a few millimetres into the tissue.

There are also other, less fundamental, but equally real limitations. Endomicroscopes have tiny fields of view, usually less than a millimetre across. This means that the operator has to hold the device incredibly steady (we’re talking tens of microns) in order to get good quality images. If the microscope shakes rapidly then there will be motion artefacts due to the raster scanning, and it is very hard to identify structures when the image is constantly moving. Trying to scan over any kind of large area in a systematic way is generally very difficult, and mosaicking techniques which stitch together different frames in order to make one big image are still not reliable.

So the good news for researchers is that, while endomicroscopy looks promising, there are still a number of engineering challenges to tackle. Some will be specific to fluorescence microscopy, while others will have application to other in vivo imaging techniques such as OCT and two-photon microscopy. And of course some may never be solved at all.


  1. Dickensheets, D.L., G.S. Kino, and L. Fellow, Silicon-Micromachined Scanning Confocal Optical Microscope. Scanning, 1998. 7: p. 38-47.
  2. Gmitro, A.F. and D. Aziz, Confocal microscopy through a fiber-optic imaging bundle. Optics Letters, 1993. 18: p. 565-567.
  3. Juskaitis, R.W., T, Real-Time White Light Reflection Confocal Microscopy Using a Fibre-Optic Bundle. Scanning, 1997. 19: p. 15-19.
  4. Sabharwal, Y.S., et al., Slit-scanning confocal microendoscope for high-resolution in vivo imaging. Applied optics, 1999. 38: p. 7133-44.
  5. Rouse, A.R. and A.F. Gmitro, Multispectral imaging with a confocal microendoscope. Optics Letters, 2000. 25: p. 1708-1710.
  6. Makhlouf, H., et al., Multispectral confocal microendoscope for in vivo and in situ imaging. Journal of biomedical optics, 2008. 13: p. 044016.
  7. Xie, T., et al., Fiber-optic-bundle-based optical coherence tomography. Optics letters, 2005. 30: p. 1803-5.
  8. Oh, W.-Y., et al., Spectrally-modulated full-field optical coherence microscopy for ultrahigh-resolution endoscopic imaging. Optics express, 2006. 14: p. 8675-84.

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>