A revolution in resolution

17 Jun 2002

Multiplexing ultrafast pulses in time improves the resolution and contrast of parallelized real-time three-dimensional microscopy. Michael Hatcher reports.

From Opto & Laser Europe April 2001

Last year Stefan Hell and his team at the Max Planck Institute for Biophysical Chemistry in Göttingen, Germany, beat the diffraction barrier in fluorescence microscopy. They developed a technique called stimulated emission depletion microscopy, which reduced the lateral extent of a confocal spot to 70 nm - the lowest value ever reported.

Now Hell - an International Commission on Optics prizewinner - is taking on real-time three-dimensional imaging. He has found a method that resolves the conflict between using parallel beams and achieving high axial resolution: crosstalk between adjacent light beams can distort an image.

The time-multiplexed multifocal multiphoton microscope (TMX-MMM) developed in collaboration with Alexander Egner and Volker Andresen could signal an end to the need for lateral scanning in real-time three-dimensional imaging, thus reducing the microscope's dependence on moving parts.

In multiphoton imaging an ultrafast laser enables a sample to absorb two infrared photons simultaneously. This is equivalent to the absorption of one ultraviolet photon.

Since two-photon absorption can only occur at the beam's focus, the technique can optically section a sample. However, a drawback with any single-beam imaging, whether it is multiphoton or confocal, is the time that is needed to obtain an image.

One remedy is to split the laser beam into between 20 and 50 sharply focused beamlets using a microlens array. This parallelization reduces image-acquisition times by the number of microlenses that are used. In Hell's instrument the array is arranged in a spiral and mounted on a rotating disk.

An ultrafast Ti:sapphire laser illuminates the array to create a number of foci, each one with a maximum allowable power of 10 mW. However, the density of foci in the sample conflicts with the achievable optical sectioning. This is because the closer the foci are to each other, the more crosstalk there is. So, for interfocal distances of less than 5 µm a haze is seen in the captured image.

Hell and his colleagues Arjan Buist and Fred Brakenhoff at the University of Amsterdam, the Netherlands, have found a way to get around this problem: separating the illuminating foci in time.

Hell's microscope forces some of the beamlets through 300 µm-thick glass. This induces a delay and temporally separates these pulses from those that do not travel through the glass. In this way interference between the 130 fs pulses of the excitation beams is eliminated.

Hell and Egner designed the delay-glass disk with patterned holes that match some of the microlenses in the array. Using a pair of glass disks means that each laser beamlet passes through one of three types of microlens, denoted A, B and C. These generate pulses with three classes of temporal delay that can then be used in closer proximity without compromising the microscope's axial sectioning.

Buist and Brakenhoff separate the illuminating foci in time by using cascaded half-silvered mirrors and beamsplitters to delay individual pulses according to which path they travel. The mirrors can easily produce a multitude of delays, but they cause an unequal split in source intensity and the beams need to be scanned galvanometrically. Hell believes that his spinning microlens array is simpler and more robust.

Hell says that, except for strongly scattering samples such as live brain slices, the TMX-MMM can achieve the same resolution as a single-beam instrument, while maintaining fast acquisition times and a high photon flux.

In theory, time multiplexing could make scanning obsolete because it allows the foci of beamlets so close to each other that their diffraction patterns overlap in space. Removing the need for lateral scanning and therefore the need for moving parts in the microscope brings obvious benefits to the user. "We could just axially scan the object through the focal plane of the microscope. The intrinsic sectioning of MMM means that only the focal plane is multiphoton-excited and sharply imaged. It's like axially scanning using the full focal plane rather than using a spot," said Hell. The applications of such a device are in biology and materials analysis. "It's suitable for extremely fast [real-time] three-dimensional imaging in a small volume," said Hell. One example is imaging calcium ions in cells, although Hell points out that such biological applications would have to take the effects of high photon fluxes on living tissue into consideration.

Replacing the rotating microlens array with a stationary set-up is also possible. "This is an intriguing idea, although it would inevitably reduce the microscope's field of view," said Hell.

Although reluctant to reveal any details, he told OLE that the TMX-MMM technique definitely has a commercial future.