A key strength of confocal imaging is the elimination of emitted light from outside the focal point, such as that from tissue autofluorescence. But there are situations when non-confocal imaging is a better option. Deconvolution or restoration microscopy, in which computational algorithms remove artefactual blurring from actual image fluorescence while restoring out-of-focus fluorescence to its proper location, is one solution.

Applied Precision of Issaquah, Washington, was the first company to offer a complete real-time restoration-microscopy instrument, the DeltaVision RT. This combines a motorized stage with deconvolution software to collect and process two-dimensional image sections for the real-time assembly of 'restored' three-dimensional image projections.

Huygens software from Scientific Volume Imaging restored the detail to this macrophage imaged by wide-field microscopy. Credit: J. EVANS

Although restoration microscopy is sometimes presented as an economical alternative to confocal, many confocal users find advantages in computational image correction, including compensation for potential resolution loss with spinning-disk instruments. Confocal manufacturers have responded by incorporating deconvolution into their software packages. For example, Nikon Instruments in Melville, New York, offers a '2D-RT decon' module that removes blurring from two-dimensional sections to clean up three-dimensional confocal images in real time, and Olympus of Tokyo uses deconvolution tools developed by Intelligent Imaging Innovations in Denver, Colorado.

Many users also opt for dedicated deconvolution software, such as the Huygens suite from Scientific Volume Imaging of Hilversum, the Netherlands. “Our software offers as much knowledge as we dare to put in about microscope image formation and noise characteristics,” explains founder Hans van der Voort, “and with that we can recover as much as possible about the original object.“

Huygens is regularly updated to tackle the data being generated by new imaging techniques — an onerous task for an expanding field. Above all, the challenge is keeping the final image clean and true-to-life. “What everybody hates is a restored image that is an artefact,” says van der Voort. “You can make an image like a photograph, which is nice to see. But at second glance, if you really want to analyse your data, what you want is reliability.”