Ultrathin multimode-fibre (MMF) holographic endo-microscopy has demonstrated cellular and subcellular imaging of neuronal structure, calcium activity and blood flow anywhere in the anaesthetized mouse brain through a bare 110 μm fibre, but its Discussion (Stibůrek et al., Nat. Commun. 2023) sets out what still stands between the technology and behaving-animal neuroscience: pixel rates orders of magnitude above the current ≈27 kHz DMD limit to support voltage indicators and larger imaged volumes, for which strategies are already being actively explored; high-quality imaging through several-hundred-millimetre fibre segments subjected to the deformations expected in experiments with freely moving animals, which the authors single out as very challenging; and chronic, reconnectable implantation allowing repeated return to the same identified circuits, which they aim to provide through custom connector solutions or implantable thin-wall guiding glass tubes. The Results additionally acknowledge one unperformed control: how much the side-view probe actually disturbs the tissue could not be assessed. Substantial component progress exists — silicon-photonic phased arrays reaching megahertz-class modulation, MEMS phase-only light modulators, ion-exchange waveguides with deformation-enduring light transport, hydrogel fibres, MMF reconnection protocols — but no endo-microscopy through a flexing fibre in a freely moving animal and no voltage-indicator imaging through any MMF endoscope have been demonstrated. The problem remains open.

Background

Deep-brain in vivo microscopy faces a fundamental access trade-off: standard two-photon microscopy reaches only ~1 mm below the cortical surface, while reaching deeper structures has traditionally required bulky gradient-index (GRIN) lens rods (typically 0.5–1 mm or more in diameter) that displace or destroy the overlying tissue. A complementary approach is holographic endo-microscopy through a bare multimode optical fibre (MMF) — a glass fibre typically ~100 μm thick whose core supports thousands of propagation modes. Because the many modes interfere, light entering the fibre is scrambled; the scrambling is described by a transmission matrix (TM), the linear map from input-field degrees of freedom to output fields at the fibre tip. Once the TM is calibrated, a spatial light modulator (SLM) — here a digital micromirror device (DMD), a chip of electronically steerable micromirrors operating at binary amplitude — can pre-shape the input wavefront so that a diffraction-limited focus appears at any chosen point at the fibre tip, enabling laser-scanning fluorescence imaging with no optics at the distal end. Stibůrek, Ondráčková, Tučková, Turtaev, Šiler, Pikálek, Jákl, Gomes, Krejčí, Kolbábková, Uhlířová and Čižmár (Nat. Commun. 14, 1897, 2023) used this principle to build an endo-microscope just 110 μm thick that resolved dendritic spines, intracellular calcium signals and single-vessel blood flow throughout the depth of the mouse brain, including via side-view fibre terminations (angled fibre tips imaging laterally) that leave the trajectory above the target largely undisturbed. Their Discussion frames what the technology still lacks, and their Results acknowledge one missing control. First, speed: the pixel rate is limited to ≈27 kHz by the DMD, whereas fast genetically encoded voltage indicators — fluorescent proteins reporting action-potential-scale membrane-voltage transients (milliseconds), as opposed to the slower calcium indicators reporting activity indirectly over tens of milliseconds — require imaging rates potentially orders of magnitude higher, as does enlarging the imaged volume; this demands faster light-modulator technology or on-the-fly computer control of modulation sequences, strategies the authors note are already being actively explored. Second, flexibility: in a freely moving animal the fibre segment outside the brain bends continuously, and each deformation changes the TM; high-quality imaging over hundreds-of-millimetres-long flexing fibre segments remains unsolved, with proposed remedies being rapid TM recalibration from pre-measured TM libraries or model-based TM computation (both requiring on-the-fly SLM control), or fibres whose light transport is intrinsically resilient to bending. Third, behavioural state: all demonstrations were acute experiments under anaesthesia, with awake head-fixed animals on treadmills in virtual reality (an experimental arrangement in which the animal runs while viewing a projected environment) as the intended next step — for which the authors regard the present technology as already suitably advanced — and freely moving animals as the ultimate goal, important because head-fixed virtual reality lacks vestibular cues and suits only a restricted behavioural repertoire. Fourth, chronicity: returning to the same neuronal circuits across sessions spanning disease progression or training requires connectorised or implanted guiding elements that keep the option of stitching adjacent fields of view into extended records, an aspiration the authors state they aim to fulfil. Fifth, tissue impact: in the Results the authors record that they could not provide the side-view equivalent of their straight-view probe-insertion blood-flow control, so the degree to which the side-view probe leaves tissue genuinely undisturbed is unquantified; translation to larger animals, including non-human primates, is untested, although the authors see no technological obstacle to it.

Problem Statement

Extend ultrathin multimode-fibre holographic endo-microscopy — as demonstrated at 110 μm probe thickness with cellular and subcellular resolution in acute, anaesthetized mouse brain — to the regime required for behaving-animal neuroscience. The objective is an instrument and methodology, built on the same bare-MMF, transmission-matrix-calibrated, distal-optics-free principle, that makes deep-brain imaging possible in awake, freely moving animals and that pursues the capability directions the source identifies: (1) imaging speed (pixel rate) increased by orders of magnitude relative to the ≈27 kHz DMD-limited rate of the source instrument, sufficient to image fast fluorescent activity sensors such as genetically encoded voltage indicators and to enlarge the observable volume; (2) preservation of diffraction-limited imaging quality while the fibre flexes during freely moving animal experiments — over fibre segments of several hundred millimetres subjected to the deformations such experiments entail; (3) chronic recording capability — connectorised or implanted guiding elements that allow repeated imaging sessions returning to the same identified neuronal circuits over extended periods while retaining the ability to stitch adjacent fields of view — within an animal-model progression running from acute anaesthetized preparations through awake head-fixed virtual-reality experiments to fully freely moving animals; and (4) assessment of how much tissue disturbance the side-view probe actually causes, which the source records as an unperformed control. What the source itself singles out as still very challenging, and what remains undemonstrated, is direction (2): maintaining high-quality imaging through a long, continuously deforming extra-cranial fibre segment in a freely moving animal, with the faster modulation and the chronic reconnectable access of directions (1) and (3) being the source's stated agenda for making that capability usable in behaving-animal neuroscience. The task is resolved by demonstrating direction (2) alone: maintaining diffraction-limited imaging quality through a long, continuously deforming extra-cranial fibre segment in a freely moving animal. Directions (1), (3) and (4) are related agenda items and need not be resolved for the task to be complete. Translation to larger animals, including non-human primates, is explicitly a subsequent extension beyond this problem, as the source itself anticipates no technological obstacle to it once the core capability exists.

References

  1. Miroslav Stibůrek, Petra Ondráčková, Tereza Tučková, Sergey Turtaev, Martin Šiler, Tomáš Pikálek, Petr Jákl, André Gomes, Jana Krejčí, Petra Kolbábková, Hana Uhlířová, Tomáš Čižmár (2023). 110 μm thin endo-microscope for deep-brain in vivo observations of neuronal connectivity, activity and blood flow dynamics. Nature Communications 14, 1897. DOI: 10.1038/s41467-023-36889-z. https://doi.org/10.1038/s41467-023-36889-z
  2. Yang Du, Evelyn Dylda, Miroslav Stibůrek, André D. Gomes, Sergey Turtaev, Janelle M. P. Pakan, Tomáš Čižmár (2024). Advancing the path to in-vivo imaging in freely moving mice via multimode-multicore fiber based holographic endoscopy. Neurophotonics 11(S1), S11506. DOI: 10.1117/1.NPh.11.S1.S11506. https://doi.org/10.1117/1.NPh.11.S1.S11506
  3. Hana Uhlířová, Miroslav Stibůrek, Tomáš Pikálek, André Gomes, Sergey Turtaev, Petra Kolbábková, Tomáš Čižmár (2024). "There's plenty of room at the bottom": deep brain imaging with holographic endo-microscopy. Neurophotonics 11(S1), S11504. DOI: 10.1117/1.NPh.11.S1.S11504. https://doi.org/10.1117/1.NPh.11.S1.S11504
  4. Lyubov V. Amitonova (2024). Multimode fiber endoscopes for computational brain imaging. Neurophotonics 11(S1), S11509. DOI: 10.1117/1.NPh.11.S1.S11509. https://doi.org/10.1117/1.NPh.11.S1.S11509
  5. José C. A. Rocha, Terry Wright, Unė G. Būtaitė, Joel Carpenter, George S. D. Gordon, David B. Phillips (2024). Fast and light-efficient wavefront shaping with a MEMS phase-only light modulator. Optics Express 32(24), 43300. DOI: 10.1364/OE.541857. https://doi.org/10.1364/OE.541857
  6. Gaolei Hu, Yue Qin, Hongnan Xu, Hon Ki Tsang (2025). High-Speed Multimode Fiber Imaging Using Binary-Modulated Silicon Photonics Optical Phased Array. Laser & Photonics Reviews 19(13), 2402206. DOI: 10.1002/lpor.202402206. https://doi.org/10.1002/lpor.202402206
  7. Hlib Kupianskyi, Simon A. R. Horsley, David B. Phillips (2024). All-optically untangling light propagation through multimode fibers. Optica 11(1), 101. DOI: 10.1364/optica.502144. https://doi.org/10.1364/optica.502144
  8. Sergey Turtaev, Tomáš Tyc, Ulf Poßner, Tina Eschrich, Torsten Poßner, Yang Du, André Gomes, Bernhard Messerschmidt, Tomáš Čižmár (2025). Deformation enduring conveyance of structured light through multimode waveguides and its exploitation for flexible hair-thin endoscopes. arXiv:2512.23514. https://arxiv.org/abs/2512.23514
  9. Tomáš Pikálek, Miroslav Stibůrek, Tereza Tučková, Petra Kolbábková, Sergey Turtaev, Jana Krejčí, Petra Ondráčková, Hana Uhlířová, Tomáš Čižmár (2025). Hair-thin confocal fluorescence endo-microscopy for deep-brain in-vivo imaging. arXiv:2512.19419. https://arxiv.org/abs/2512.19419
  10. Simon P. Mekhail, Hao-Wei Hu, Elina Köster, Graham Gibson, Miles J. Padgett (2026). Portable system for flexible micro-endoscopy using a multi-mode fiber. Advanced Photonics Nexus 5(2), 026011. DOI: 10.1117/1.APN.5.2.026011. https://doi.org/10.1117/1.APN.5.2.026011
  11. Peng Chen, Keyi Han, Zijun Gao, Claire M. Deng, Haoyu Xu, Zhi Ling, Corey Zheng, Mithila Sawant, Marcus T. Cicerone, Aparna H. Kesarwala, Jeffrey E. Markowitz, Shu Jia (2026). Hydrogel Fiber Endomicroscopy. bioRxiv preprint. DOI: 10.64898/2026.03.23.713710. https://doi.org/10.64898/2026.03.23.713710
  12. Le Yang (2025). Mismatch reconstruction theory for unknown measurement matrix in imaging through multimode fiber bending. arXiv:2510.21787. https://arxiv.org/abs/2510.21787