Imaging inside scattering media

When light propagates through a disordered medium, it gets scattered numerous times, which scrambles the optical wavefront. This is why biological tissue, bone, colloidal suspension, clouds, and fogs are opaque. Despite its complexity, this multiple scattering process is reversible thanks to the time-reversal symmetry of Maxwell’s equations. One goal of our group is to leverage this property and “unscramble” light scattering to see through opacity.

By time-reversing the light emitted from a point source (i.e., a guidestar), one can focus light back to the originating source location despite multiple scattering along the path. Scanning the location of the guidestar and repeating the time-reversal operation can, in principle, create a high-resolution image deep inside a scattering medium. However, the requirement of such guidestars severely restricts the applicability of such a scheme. Existing guidestar-free wavefront shaping methods have other limitations such as only working for planar targets outside the scattering medium and requiring fluorescence labeling or transmission measurements. Adaptive optics and existing matrix-based methods, in the meanwhile, cannot correct for strong scattering.

We develop an approach called “scattering matrix tomography” (SMT) [1,2] that converts the problem of imaging and reversing light scattering to a computational reconstruction and optimization problem, as illustrated in the figure below. The scattering matrix element S(k_out, k_in, ω) is the outgoing amplitude in direction k_out given an incident wave from direction k_in at frequency ω. We measure the scattering matrix of the sample, use it to digitally scan a synthesized confocal spatiotemporal focus and construct a coarse volumetric image of the sample, and then use the synthesized image as many virtual guidestars to digitally optimize the dispersion compensation θ(ω), input wavefront ϕ_in(k_in), and output wavefront ϕ_out(k_out) to compensate for aberrations and scattering. The virtual feedback dispenses with physical guidestars and enables hardware-free spatiotemporal wavefront corrections across arbitrarily many isoplanatic patches. SMT is noninvasive and label-free, and it provides volumetric images inside and outside the scattering media with unprecedented depth-over-resolution ratios. It can be applied to medical imaging, device inspection, biological science, and colloidal physics.

Related publications

1. Deep imaging inside scattering media through virtual spatiotemporal wavefront shaping, Yiwen Zhang, Minh Dinh, Zeyu Wang, Tianhao Zhang, Tianhang Chen, and Chia Wei Hsu. arXiv:2306.08793.
2. Full-wave simulations of tomographic optical imaging inside scattering media, Zeyu Wang, Yiwen Zhang, and Chia Wei Hsu. arXiv:2308.07244.