Funnelling super-resolution STED microscopy through multimode fibres

Abstract

Holographic multimode fibre endoscopes have recently shown their ability to unveil and monitor deep brain structures with sub-micrometre resolution, establishing themselves as a minimally-invasive technology with promising applications in neurobiology. In this approach, holographic control of the input light field entering the multimode fibres is achieved by means of wavefront shaping, usually treating the fibre as a complex medium. In contrast to other unpredictable and highly scattering complex media, multimode fibres feature symmetries and strong correlations between their input and output fields. Both step-index and graded-index multimode fibres offer a specific set of such correlations which, when appropriately leveraged, enable generating high-quality focused pulses with minimal intermodal dispersion. With this, we funnelled pulsed super-resolution STED microscopy with time-gated detection through a custom multimode fibre probe, combining the correlations of both multimode fibre types. We demonstrate resolution improvements over 3-times beyond the diffraction limit and showcase its applicability in bioimaging. This work provides not only a solution for delivering short pulses through step-index multimode fibre segments but also marks a step towards bringing advanced super-resolution imaging techniques with virtually no depth limitations.

Figures (15)

• Figure 1: Focussing and shaping pulsed laser light through an endcap multimode fibre (MMF) endoscope.a, Illustration of the endcap probe working principle. Light from an annular region couples and propagates through a step-index MMF segment mainly within a single mode group (with conserved propagation angle, $\alpha$), being focussed at the output by the graded-index MMF endcap. b-d, Experimental axial extent of a 485nm pulsed laser focussed through a 105µm-diameter core and $0.22$ NA step-index multimode fibre, a 100µm-diameter core and $0.29$ NA graded-index multimode fibre, and a 100µm-diameter core and $0.29$ effective NA endcap fibre probe, respectively. e-g, Experimental axial extent of a 592nm pulsed laser vortex beam focussed through the step-index, graded-index, and endcap fibre probes, respectively. h-j Experimental temporal profile of a 485nm (blue) and a 592nm (orange) picosecond pulsed lasers focussed through the step-index, graded-index, and endcap fibre probes, respectively. k-m High dynamic range images, in logarithmic scale, of the pulsed excitation (diffraction-limited focus) and depletion (vortex beam) lasers at the output of the respective different fibres as mentioned and labelled in b-g. The inset in k shows the location of the generated beams in respect to the fibre core (circle line). In all cases, the beams were created by controlling a single circular input polarisation and by employing phase-only modulation. Scale bar: 10µm.
• Figure 2: Performance of STED microscopy using pulsed lasers through an endcap fibre probe.a, Diffraction-limited image and b, STED image of a 400nm fluorescent nanosphere. c, Diffraction-limited image and d, STED image of a 400nm fluorescent nanosphere cluster. e, FWHM dependence on the average STED beam power determined from imaging 500nm and 400nm fluorescent nanospheres. Each datapoint and corresponding error bar is the result from an average between 10 images of distinct fluorescent nanospheres. The curve for 0nm outlines the projection of the fitting model to an infinitely small fluorescent centre, indicating an effective point-spread function FWHM close to 200nm. f, Example of a time-correlated single-photon counting histogram from a 500nm fluorescent nanosphere measured at two indicated STED beam powers ($0$ and 107mW). The time-gate detection window starts after the excitation and depletion events, avoiding counting photons not yet depleted. g Normalised cross-section of a and b (vertical and horizontal average) demonstrating a resolution improvement of about 3-fold over the diffraction-limit. h, Normalised cross-section corresponding to the dashed lines in c and d, revealing 4 resolved spheres in the STED image (limited by the sphere size). Average excitation power in a-d: 2.4µW, e: 2.4µW for 400nm spheres and 0.6µW for 500nm spheres; STED beam average power in b, d: 107mW. Dwell time: 1ms. Scale bars: 500nm.
• Figure 3: STED imaging of cells through an endcap fibre endoscope.a, Panel consisting of several diffraction-limited images of a HEK293 cell, in vitro, with actin stained with Alexa Fluor 488 Phalloidin. b, Diffraction-limited image and c, STED image of the dashed blue line region in a. d, Diffraction-limited image of peroxisomes in a HEK293 cell, in vitro, with sterol carrier protein 2 (SCP2) stained with enhanced green fluorescent protein (eGFP). e, Diffraction-limited image and f, STED image of the dashed blue line region in d. g Normalised cross-section of the arrow-marked regions in e and f, showing three distinguishable peroxisomes in the STED image, which are not resolved in the diffraction-limited image. Average excitation power in a-c: 1.5µW, d-f: 1.0µW; STED beam average power in c, f: 56.4mW. Dwell time: 1ms. Scale bars: 2µm.
• Figure S1: Axial extent comparison between the pulsed excitation beam and pulsed depletion beam (vortex beam for 2D-STED and bottle beam for z-STED).a, Excitation beam, d, depletion vortex beam, and g, depletion bottle beam axial extent for a 105µm core and 0.22 NA step-index multimode fibre. b, Excitation beam, e, depletion vortex beam, and h, depletion bottle beam axial extent for a 100µm core and 0.29 NA graded-index multimode fibre. c, Excitation beam, f, depletion vortex beam, and i, depletion bottle beam axial extent for a 100µm core and 0.29 effective NA endcap multimode fibre. The calibrated focal plane is located at $Z = 50µm$.
• Figure S2: Comparison of a pulsed excitation beam through a step-index multimode fibre, in which the reference arm is tuned for maximum interference with the signal from a, lower order modes (OM), b, intermediate order modes, and c, higher order modes.