Study of invariant properties and correlations in MUltimode Fibers For ImagiNg – MUFFIN

Optical multimode fibers have become the subject of numerous investigations, in particular for applications in telecommunications and endoscopic imaging. Compared to fiber bundles or multicore fibers used in endoscopic imaging, they offer orders of magnitude higher information density, allowing us, in theory, to both increase the resolution and decrease the footprint of the device. However, dispersion and mode coupling scramble the input image, even in a few-centimeter fiber. Image reconstruction techniques through multimode fibers rely on measuring the transmission matrix, which fully characterizes the optical system. Nevertheless, this matrix typically changes in real-time due to the dynamic bending of the fiber, preventing the direct use of a previously calibrated system.

We propose in the MUFFIN project to study invariant properties that can be used to recover information without having to fully characterize the system, i.e. without having to measure its transmission matrix. In particular, the existence of a rotational memory effect has recently been observed but has not yet been theoretically studied, and its properties are not predicted. This effect holds great promise for imaging that we intend to exploit.

The goal of the MUFFIN project is twofold: first, to experimentally, theoretically, and numerically study the invariant properties of multimode fibers, and then to harness those effects for endoscopic imaging. Specifically, we will predict and explain the properties of the rotational memory effect in multimode fibers, explore new memory effects or invariant properties that can be exploited for image reconstruction, and then experimentally demonstrate their use for imaging.