Computational image enhancement of multimode fibre-based holographic endo-microscopy: harnessing the muddy modes

Abstract

In imaging geometries, which employ wavefront-shaping to control the light transport through a multi-mode optical fibre (MMF), this terminal hair-thin optical component acts as a minimally invasive objective lens, enabling high resolution laser-scanning fluorescence microscopy inside living tissues at depths hardly accessible by any other light-based technique. Even in the most advanced systems, the diffraction-limited foci scanning the object across the focal plane are contaminated by a stray optical signal carrying typically few tens of % of the total optical power. The stray illumination takes the shape of a randomised but reproducible speckle, and is unique for each position of the focus. We experimentally demonstrate that the performance of imaging a fluorescent object can be significantly improved, when resulting images are computationally post-processed, utilising records of intensities of all speckle-contaminated foci used in the imaging procedure. We present two algorithms based on a regularised iterative inversion and regularised direct pseudo-inversion respectively which lead to enhancement of the image contrast and resolution. (C) 2021 Optical Society of America under the terms of the OSA Open Access Publishing AgreementIn imaging geometries, which employ wavefront-shaping to control the light transport through a multi-mode optical fibre (MMF), this terminal hair-thin optical component acts as a minimally invasive objective lens, enabling high resolution laser-scanning fluorescence microscopy inside living tissues at depths hardly accessible by any other light-based technique. Even in the most advanced systems, the diffraction-limited foci scanning the object across the focal plane are contaminated by a stray optical signal carrying typically few tens of % of the total optical power. The stray illumination takes the shape of a randomised but reproducible speckle, and is unique for each position of the focus. We experimentally demonstrate that the performance of imaging a fluorescent object can be significantly improved, when resulting images are computationally post-processed, utilising records of intensities of all speckle-contaminated foci used in the imaging procedure. We present two algorithms based on a regularised iterative inversion and regularised direct pseudo-inversion respectively which lead to enhancement of the image contrast and resolution. (C) 2021 Optical Society of America under the terms of the OSA Open Access Publishing Agreement