Conference Agenda

Holographic multimode fibre endoscopes have established themselves as a tool for minimally invasive imaging, with particularly promising applications in the domains of neurobiology. These instruments allow imaging of previously inaccessible deep brain regions of living animals models. In all these applications, wavefront shaping is used to holographically control the input light fields entering the multimode fibre, with it being treated purely as a complex medium. Yet, multimode fibres exhibit symmetries and strong input-output field correlations, which are distinct for step-index and graded-index multimode fibres. In this work, we appropriately leverage these correlations to enable high-quality focussing of pulsed lasers with minimal intermodal dispersion. Such effect is then used as the underlaying basis for delivering pulsed super-resolution STED microscopy through a custom multimode fibre endoscope comprising input-output correlations of both step-index and graded-index fibre types. We show resolution improvements over 3-fold beyond the diffraction limit and showcase its applicability to bio-imaging. This work offers a solution for delivering short pulses through step-index segments and represents a step towards enabling advanced imaging techniques with virtually no depth limitation.

Lensless endoscopy enables minimally invasive imaging of biological tissues, particularly in the brain. However, miniaturization and optical performance remain key challenges. Multicore fibers (MCFs) are promising probes, where beam shaping at the output is typically achieved using Spatial Light Modulators (SLMs) at the input. Even though SLMs are active components, they require bulky optical elements and have efficiency and speed limitations.

Metasurfaces offer a compact alternative for wavefront shaping. By integrating them with MCFs, we aim to create a highly flexible two-photon endoscope. Metasurfaces can perform phase and group delay compensation with improved resolution while significantly reducing system footprint, as they match the fiber’s dimensions.

We characterize fiber transmission and metasurface performance. First, we measure the fiber’s transmission matrix, determine the required curvature for focusing, and fabricate the metasurface via lithography and etching. After fabrication, phase and transmission measurements validate its optical response. The metasurface is then imaged at the fiber input to realise the focusing at the output.

Our demonstration features a scanning endoscope using a passive metasurface for phase compensa-tion and focal spot generation, with beam scanning controlled by galvo mirrors. Finally, we discuss future prospects, including group delay compensation and active metasurfaces for dynamic wave-front control.

We propose a new spectroscopy technique in the mid-infrared (MIR) domain without any cryogenic system. The MIR field emitted by the source is shifted to the near-infrared using sum-frequency generation in a Periodically Poled Lithium Niobate ridge waveguide, and is then detected in the photon-counting regime using a low-noise SiAPD detector. The non-linear process is powered by a continuously tunable pump laser. We present here an experimental proof-of-concept, where the light emitted by a thermal source in the 3-4μm band is spectrally modulated using a Michelson interferometer. By continuously tuning the pump laser wavelength between 1058nm and 1078nm, the MIR spectrum is reconstructed, and the imposed spectral modulation period is retrieved successfully with a relative error of less than 5%

In this work, we present a novel approach to determine the refractive index n and the extinction coefficient κ from terahertz time-domain spectroscopy (THz-TDS) measurements using artificial neural networks (ANNs). We train the network in our experimental datasets to create a model that can accurately and efficiently calculate these material properties, the proposed neural network is superior to traditional analytical methods, which rely on approximations, and a faster and easier alternative to iterative root-finding algorithms. Our results demonstrate that this machine learning methodology solves common issues in THz-TDS data analysis, such as phase unwrapping, time-domain windowing, low computation rates, and high accuracy at low-frequency regions, effectively and achieve results with high accuracy.