Sharpening the focus of optical time-reversal in foggy media

Imaging in biological tissues can be likened to focusing light in fog, where optical scattering sends the ray along meandering trajectories. Barring an astronomically unlikely coincidence, it is hard to imagine that the rays would somehow spontaneously converge to create an intense spot of light. Biological tissues are essentially foggy in nature. In fact, we appear opaque because of the large amount of deviation that occurs in light's trajectory when we direct it through tissue.

Scattering has traditionally been seen as a problem that scrambles information, but it is actually a deterministic and reversible process.^1–4 Therefore, there exists a wavefront solution for directing light into a scattering medium so that the field would generate a speckle-size limited focus spot. The big challenge is finding that wavefront without direct optical access to the inside of the tissue.

Professor Lihong Wang's group at Washington University at Saint Louis recently demonstrated a technique—time-reversal of ultrasound-encoded light (TRUE)—that uses an acousto-optic guidestar (an artificial means of correcting distortion) for optical time reversal, which alters the propagation direction and phase variation of the light beam, sending it in the exact opposite direction. The method achieves focusing in scattering tissue phantoms and ex vivo tissues.⁵ Wang's group used ultrasound, since the waves are not significantly scattered in soft biological tissues, and can be focused at significant depths. A portion of the scattered light that passes through the ultrasound focus shifts in frequency, resulting in an acousto-optical ‘beacon’ within the tissue. This enables detection and time-reversal of the frequency-shifted light emanating from the beacon to achieve optical focusing at the location of the ultrasound focus.

While the original demonstration of TRUE showed sub-millimeter-scale imaging with absorption contrast, the immediate application of this technique for fluorescence imaging remained a challenge because of the low optical gain of the photorefractive crystal-based time-reversal mirror. To address this, we demonstrated high-resolution (∼40μm) TRUE focusing and focal fluorescence imaging in deep tissues using an optoelectronically-implemented time-reversal mirror—termed a digital optical phase conjugator (DOPC)—to enable high optical gain.⁶ With this new flexibility, we were able to push time-reversal optical focusing even closer to its fundamental focusing limit.

To understand what that limit is, we can revisit our prior assertion about the existence of an optimal wavefront that would create a speckle-size limited focus within the scattering medium. Moreover, every point in the medium has such a wavefront. In TRUE, the focused spot size is determined by the ultrasound spot size (tens of μm at best). This is because the solution found with TRUE is a sum of optimal wavefront solutions for all spots within the ultrasound focus.

To overcome this resolution limitation and ultimately achieve single optical speckle resolution imaging, we have developed a new technique: time-reversal of variance-encoded light (TROVE).⁷ Here, we encode the position of speckles at much finer resolution than dictated by the size of the ultrasound focus (see Figure 1). We illuminate the sample with a sequence of random optical waves and measure the series of ultrasound-tagged wavefronts that emerges. Each is composed of a mixture of the optimal wavefronts associated with points in the ultrasound focus. Assuming that the recorded sequence is long enough, we can then tease out each wavefront, the most prevalent in the sequence being associated with light emerging from the center of the ultrasound focus (because that is the point where light is most strongly modulated). As such, when we time-reverse this wavefront, it converges back to the exact center of the focus. This, in essence, enables us to super resolve the focal size and achieve optical resolution with diffuse light. We can further manipulate the data taken with three or more different focus positions to super resolve to several points in the medium.

Figure 1. Rendering of an ultrasound frequency-shifted optical speckle field. Color represents phase and luminance represents amplitude.

Despite their promise of deep tissue fluorescence imaging, much remains to be done for TRUE and TROVE to be applicable in biological tissues in vivo. The techniques require acquisition times to be shorter than the timescale of mechanical tissue fluctuations (in the order of milliseconds), and need control of wavefronts with higher acuity and resolution. However, with technical improvements on the horizon, these methods could in the future provide a wide range of imaging and photostimulation applications at unprecedented depths.