Deep-brain fluorescence imaging

Much of what we know about the brain, we know because we can image it using a microscope. We can look at the tiny features inside each neuron to see how it is built, or look at the electrical activity by using dyes which glow whenever parts of the brain are working. Unfortunately, this means that we know much more about the surface of the brain than we do about the parts buried deep within, just because we can't squeeze a large microscope into such a small space. To truly understand the brain, we must find a way to image the buried parts too.

Putting a microscope inside the brain is hard, because lenses are big. Previously, people have tried to image using tiny microscopes, but while it is possible to get an image quite quickly using these microscopes, they are bulky and do considerable damage to the brain as they are inserted. An optical fiber can get the same image, but with a much smaller diameter, which means less damage. Understandably, if we cause too much damage, we can no longer be certain that what we observe in the brain reflects how it acts in its natural state; a situation which is largely the current state of high-resolution deep brain imaging today.

Unfortunately, the light that travels along an optical fiber is scrambled. This project is focused on finding a way to unscramble it, using holograms - patterns of light that encode all the information in an image, rather than just part of it as one would see in a 2D photograph. By projecting a hologram instead of a photograph, we can 'pre-scramble' the light, compensating for the effect of the fiber and allowing us to image deep inside the brain. We will design and build a machine that can make holograms extremely quickly - around 25 thousand of them per second. By carefully controlling them, we can project images that appear scrambled, but once they travel down the optical fiber, they turn into the pattern of light that we want.

The instrument we would like to make is based on the same type of chip that is used in some digital projectors, consisting of tiny little mirrors that flip back and forth very rapidly. This lets them delay the light hitting them by a very small amount, which when all the pixels operate together, allows us to unscramble the light travelling through the optical fiber. Previously, we have shown that we can project holograms using a different technology, but this technology is very slow. In contrast, the new technique described here can image just as fast as a normal microscope, which means we can see more features in the brain, especially things that change, like electrical activity.

The holographic projector we will develop is useful for many other applications - 3D printing, holographic TV, as well as controlling a mouse brain by switching neurons on and off. It is important because it is approximately one hundred times faster than other devices that can do similar things, so even if it doesn't work in the mouse brain, it will be very useful in other areas.

Because cortical regions are accessible to optical microscopy and patch-clamp techniques, far more is known about them than the subcortical regions. Despite much research into non-invasive methods, only the insertion of optical lens systems into the brain has permitted investigation of multiple subcortical neurons with subcellular resolution. Since the resulting damage is extensive though, significant doubt is cast on the validity of any conclusions drawn from such models. We seek to replace these lenses with optical fibers, drastically reducing the resulting tissue displacement; the use of a dispersed array of these fibers means that large areas of the brain can still be imaged, despite the low displaced volume.

We will develop a wavefront correcting element that can operate at above 10kHz, and use it to correct for heavy aberration in a multimode optical fiber. By projecting multiple different phase patterns onto the proximal end of the fiber, a scanning focal spot can be projected onto the distal end. Multiphoton fluorescence can then be captured by the same fiber and detected using a photomultiplier tube. Once this device is complete, it will be combined with a short multimode fiber consisting of a glass fiber coated with protected silver, and used to image subcortical neural activity in mice expressing a fluorescent calcium indicator.

The use of this instrument to image subcortical activity is but the tip of the iceberg when it comes to potential applications. The same instrument can also be used to overcome tissue scattering when imaging into organs such as the brain without the use of implants, or to perform high-throughput holographic patterning of photopolymers for 3D printing. Applications also exist in selective optogenetic excitation of large numbers of neurons simultaneously, holographic displays, and high-precision targeted photodynamic therapy.

A large number of people stand to gain from this research. Initially, neurobiologists who wish to see below the outer layers of the brain stand to gain a lot - currently they can either image using techniques like fMRI or PET, which cannot resolve individual cells, or they can use optical microscopy by either surgically removing the outer layers of the brain, or poking a small microscope objective into it. Our technology allows neurobiologists to see things like neural activity by imaging down a very thin optical fiber at high speeds.

The devices developed in this instrument are applicable elsewhere as well. Biologists stand to gain from a wavefront shaping element that can project arbitrary light patterns onto a surface, for uses in tissue ablation, photodynamic therapy and optogenetics. In all these cases, current technology is limited, in the sense that there are no photon-efficient ways to project arbitrary patterns at high speed. Biologists seeking to image without being limited by tissue scattering can also employ wavefront shaping techniques to 'undo' the effect of tissue scattering, substantially increasing the optical penetration depth; currently these techniques are limited to static samples as the wavefront correction cannot be updated fast enough to overcome the ~1ms tissue decorrelation time. With this new phase modulator that limit may be overcome.

Engineers stand to benefit from the availability of such a fast wavefront shaping element, since it can be used for holographic projection in 3D printing, laser displays, free-space optical communication and more. Existing technologies are frequently slow, with display rates on the order of 60-100fps, which is slow for 3D printing, insufficiently fast to average out speckle in display applications, and not fast enough to cancel out atmospheric distortions in communications applications.

The relatively low cost of the modulator also has benefits to users of adaptive optics mirrors, which can be extremely expensive as the number of actuators increases. While it may not be sufficient for astronomical applications, it may prove suitable for other fields such as laser power projection where the image constraints are not as pronounced.