Microscopy Home PageReal objects are three-dimensional, even thin sections have a thickness although for practical purposes this is often ignored. Images are usually captured, stored and displayed in just two dimensions, and microscopy is no exception to this rule.
When it is important to capture or display three-dimensional data we have to rely on special methods which are often costly or time-consuming. Some of these methods are unique to microscopy, but others are widely used in other fields such as astronomy, architecture, engineering and geology.
At the simplest level a hand lens provides immediate information about the three-dimensional structure of any small object. Because the object can be rotated in the field of view an everyday sense of the three-dimensional structure is readily available. Practical magnifications are limited to below 15 times, between 5 and 10 times is normal.
A white light stereomicroscope gives similar information at magnifications up to about 200 fold, beyond this, depth of field and resolution are unlikely to be adequate. The scanning electron microscope (SEM) and scanning probe microscope produce images which can often be interpreted in the same way, though some care is needed.
In special cases solid specimens may be cleared by infiltrating them with a fluid of suitable refractive index. This is sometimes done with botanical materials for example. In these cases internal structures can be viewed directly as above.
Any form of microscopy can be used for this. The image may be examined visually, or captured photographically or by imaging chip (CCD), the three-dimensional object is recorded as a two-dimensional projection onto a plane.
Optical sectioning is a non-mechanical way of gathering the same kind of information, by forming an image of a thin zone inside a more or less transparent sample. A particularly effective approach for this is confocal imaging (see below).
Stereoscopic pairs of images are sometimes taken as these can provide an illusion of depth. At Cambridge University, Bernie Breton has developed an exciting real-time stereoscopic imaging system for the SEM.
However obtained, stereo pairs can be viewed in various ways. They are often presented as anaglyphs, either red/green or red/cyan, and viewed through coloured spectacles, see for example Gmmc's page at the University of East Anglia, or Bernie Breton's page already mentioned.
Alternatively prisms and/or lens systems can be used to present the correct image to each eye.
Many specimens are opaque. Unless X-rays or other penetrating beams are used the only way to see inside is to cut the sample open. Sections are another kind of two-dimensional projection, revealing internal structure. A series of sections could obviously be stacked together to recreate the original object. In the same way, photographic or other magnified images of the sections can be stacked to create a magnified three-dimensional model of the specimen.
Solid models have been constructed from transparent plastic, wood, expanded polystrene and other materials. Producing such models is time-consuming and difficult and the results are of limited flexibility and value.
With the development of powerful computers it became possible to make virtual reconstructions and display the results immediately. The images can be scanned or digitised and the reconstruction viewed from any angle, coloured, shaded and textured.
Göteborg University has examples of serial section reconstruction by computer, as does Glasgow University's Institute of Biomedical and Life Sciences.
The Ross software available from NASA is unfortunately licenced only for use in the USA, but other software is available much more widely. The BOB program from the University of Minnesota takes a 3-d array of data as input and then displays it in the form of rendered graphics. This is a downloadable UNIX package. Rendered solid images are also provided by the Volpack software library from Stanford University.
A confocal microscope uses clever optical techniques to illuminate and image only a thin plane through the specimen. Since light from other planes is rejected, most of the out-of-focus clutter is avoided leading to clear, sharp images of the region of interest. This leads to two-dimensional images which are excellent optical sections. As described earlier these can be collected and recombined to give true three-dimensional information.
This topic is particularly well-served on the Web. Lance Ladic at the University of British Columbia runs a site devoted to 3-D laser confocal microscopy. Many useful sites are listed here as well as plenty of other valuable information. Molecular Dynamics is another site offering a good deal of information on confocal instruments and methods.
This term is commonly used of medical imaging technology but can apply to microscopic data collection and processing too. The team at CSC in Finland have produced interesting results in this way.
Tomography makes a virtual three-dimensional model of the specimen and then displays virtual slices through it at any required position and angle. Although not microscopical, the famous visible human project used real sections to create a database for tomographic reconstructions.
The Lawrence Livermore National Laboratory has a page devoted to X-ray tomographic microscopy.