Creating order from disorder
ASHLEY N. RICE, ASHLEY.RICE@PHOTONICS.COM
ENSCHEDE, Netherlands, and FLORENCE, Italy – Light scattering makes clear imaging through an opaque material seem insurmountable, but researchers have discovered that the resulting speckle pattern actually contains the key to restoring the shape of the original object. The findings could have important applications in bioimaging.
The speckle pattern “is not completely random but contains some subtle correlations. We realized that the knowledge that these correlations are present was enough to get some information on the object,” Dr. Jacopo Bertolotti of the University of Twente’s MESA+ Institute for Nanotechnology told BioPhotonics
Bertolotti and colleagues from the universities of Twente and Florence took this knowledge and developed a technique that can sort randomly scattered light from light that has not changed direction to reconstruct an image – and it doesn’t require a detector to be placed behind the layer. The work was published in Nature
A cartoon by Clive Goddard, inspired by research at the
universities of Twente and Florence on a noninvasive method for
nscrambling scatted light. Courtesy of Clive Goddard ©firstname.lastname@example.org.
Because the exact shape of the speckle pattern isn’t known, the scientists could not measure the object directly, “but, as it turned out, we could measure its autocorrelation,” Bertolotti said. “And inverting the autocorrelation to give the shape of the original object is a numerical problem that computers are very good at solving.”
The angle of a laser beam illuminating an opaque diffuser was scanned. At the same time, a computer recorded the amount of fluorescent light that was returned by a tiny object hidden behind the diffuser. The computer program initially guessed the missing information, then tested and refined the guess.
“Our method is based on the speckle correlation known as the optical memory effect that effectively allows us to scan rigidly an unknown speckle pattern over the hidden object,” he said. “That would allow us to measure a 3-D autocorrelation that can be inverted to yield the original object.”
A graphical illustration of the method described in the paper. (a)
The test object used was the Greek letter p, written in fluorescent ink
and 100 times smaller than the one printed here. The test object was
covered by a strongly scattering ground-glass diffuser that hid it from
view. (b) A laser beam scanned the ground glass. The test object yielded
only a diffuse glow of fluorescent light. (c) The intensity of the
fluorescence was measured versus the angle of the laser beam and
recorded by a computer. The seemingly random pattern bears no
resemblance to the test object. (d) The computer searched for
similarities in the measured pattern, which it used to calculate the
true shape of the object. Courtesy of Dr. Elbert van Putten.
Light signals were unscrambled to retrieve detailed images of human-cell-sized (around 50 µm across) fluorescent objects hidden 6 mm behind scattering layers as well as a complex biological sample enclosed between two opaque screens.
When asked if it is possible to retrieve objects deeper than 6 mm, Bertolotti said that increasing the distance is not a problem – in fact, increasing the distance would make it conceptually easier.
“Going farther and farther away would just make the available signal fainter and, thus, would require a longer measurement,” he said. That holds true if the object is behind the layer and not immersed in a strongly scattering medium.
The method is not yet fast enough to examine objects under the skin.
“This method is intrinsically sequential,” Bertolotti said. “Each point has to be measured one after the other just like what happens with confocal microscopy. Right now, the bottleneck, in terms of time required, is the integration time needed to detect enough signal for each point. Decreasing this time will speed up the measurement.”
The team next intends to “investigate how to generalize our method to other contrast mechanisms than just fluorescence,” Bertolotti said. “Our method should work also using second-harmonic generation, optoacoustic effect, CARS [coherent anti-Stokes Raman spectroscopy] and many other mechanisms.”
The idea that order can be created out of disorder, which is “not an insurmountable obstacle to eliminate at all costs” is one that is rapidly gaining ground, Bertolotti said. “We recently demonstrated how a completely disordered layer of semiconductor can be used for high-resolution microscopy, but we are not the only ones going in this direction,” he said, citing 2010 papers by the groups of Peter Lodahl and Sylvain Gigan.