Recently, our group has introduced Dynamic Adaptive Scattering Compensation Holography (DASH), a sensorless wavefront correction method for nonlinear imaging, which allows retrieving even strong wavefront aberrations in short times.The possibility to compensate the adverse effects of high spatial frequency volumetric scatterers shifts yet another problem into the focus: the small size of corrected sample regions, which sometimes measure only a few µm across. Real-world biological applications, however, demand the imaging of larger fields, making it highly important to find ways of enlarging these correction zones. We present a sample-conjugate version of DASH, which addresses this problem in two ways: While sample-conjugation is naturally better suited to compensate the effect of certain scatterers, the use of a large pixel count SLM further allows applying multiple correction patterns simultaneously.

Aberrations are a common problem in microscopes resulting in compromised imaging contrast and resolution. Adaptive optics (AO) can correct aberrations but requires either a wavefront sensor or a wavefront-sensorless AO method that requires multiple sample exposures. We created a machine learning (ML) approach that embeds physical understanding of the imaging process into a sensorless AO method. This enables correction of aberrations with as few as two sample exposures. The method was translated across different microscope modalities. This includes two-photon microscopy and three-photon microscopy of in vivo mouse neural activity, showing robustness to specimen motion and activity related intensity variations.

Wavefront sensing, often operated using an array of microlenses, is usually considered as a low-definition imaging technique. Yet, there exists a less-known, high-definition wavefront sensing technique, based on the use of a 2D-grating (aka cross-grating) instead of a microlens array, that enables the imaging of a wavefront profile with a 0.1 nm sensitivity and a diffraction-limited resolution. When implemented on a microscope, this technique can compete with the most advanced quantitative phase microscopies. In this contribution, we will show how we used cross-grating wavefront sensing for the imaging and characterization of nanoparticles, mammalian cells, bacteria and neurites, in vitro.

The inhomogeneous refractive indices of tissues blur and distort single molecule emission patterns generating image artifacts and decreasing the achievable resolution of single molecule localization microscopy (SMLM). We developed deep learning driven adaptive optics (DL-AO) that monitors the individual emission patterns from SMLM experiments, infers their shared wavefront distortion, and drives a deformable mirror to compensate sample induced aberrations. We demonstrated that DL-AO compensates 28 types of wavefront deformation shapes, restores single molecule emission patterns approaching the conditions untouched by the specimen, and improves the resolution and fidelity of 3D SMLM through brain tissues over 130 µm, with 3-20 mirror changes.

In single-photon confocal fluorescence microscopy, sample-induced optical aberrations affect both excitation and detection paths and can lead to severely degraded image quality. We applied an adaptive optics (AO) method based on frequency-multiplexed aberration measurement to a confocal microscope. We validated this method by performing AO correction on features of different sizes with artificial aberrations and demonstrating drastic improvement in image signal, contrast, and resolution. Moreover, we showed that this method can significantly enhance imaging performance through living zebrafish larvae and in the mouse brain in vivo.

In this work, we demonstrate a method that can focus on multiple single guidestar through scattering medium with a spatial light modulator (SLM) and a bucket detector using two-photon fluorescence signals. A gradient descent based multiplexed phase retrieval algorithm is used to non-invasively reconstruct the transmission matrix between the guidestars and the SLM, without any assumptions on the memory effect range. Conversely, if we consider the memory effect, we can reconstruct the image of the sample.

We couple the T-matrix method with a discrete particle representation of turbid media to simulate the focusing light through a highly scattering titanium dioxide phantom. We have used our method to simulate wavefront shaping with full phase modulation using a stepwise sequential algorithm, and have generated multiple foci and compared their enhancement against theory. Our computationally efficient, yet physically realistic technique, allows researchers to resolve both amplitude and phase information at arbitrary locations inside and outside bespoke scattering media.

All-optical manipulation and recording of neural ensembles offer a non-invasive way to study the causal relationship between precise features of neural activity and specific aspects of sensation, cognition, and action. However, tissue scattering reduces the precision of two-photon multi-site photostimulation and limits the accessible depth. To correct multiple scattering, we propose to develop a new technique based on light field control for photostimulation and imaging of neural activity deep inside of the mouse brain.

Coherent imaging through optical multimode fiber has the potential to enable label-free imaging through hair-thin imaging probes deep inside tissues and organs. Towards this goal, we developed a method to efficiently measure the multispectral transmission matrix of multimode fiber. Combined with a computational approach that circumvents the need for active wavefront shaping, this enabled tomographic imaging with confocal and coherence-gating through multimode fiber. Together with the development of retrieving the transmission matrix of dynamically bent fiber and compressing the measurements required for image reconstruction, these efforts address critical challenges towards the realization of practical multimode fiber imaging.

The lensless endoscope represents the ultimate limit in miniaturization of imaging tools: an image can be transmitted through an optical fiber by numerical or physical inversion of the fiber's pre- measured transmission matrix. We present here a novel fiber-optic component, a "tapered multi-core fiber (MCF)", designed for integration into ultra-miniaturized endoscopes for minimally invasive two-photon point-scanning imaging. This new design addresses the power delivery issue that has faced MCF based lensless endoscopes. We achieve experimentally a factor 6.0 increase in two-photon signal yield while keeping the ability to point-scan by the memory effect. We report two-photon fluorescent imaging of cells and neurons with these improved MCF tapered fibers.

Coherent fiber bundles used in endoscopic imaging suffer from inter-core dispersion resulting in pseudo random phase distortions for a transmitted wavefront. This limits their application to relaying intensity patterns for pixelated, 2D near field imaging. In the last years, employing spatial light modulators for digital optical conjugation of these distortions and unpixellated 3D raster scanning has been demonstrated. Here we present using 2-Photon Polymerization for writing phase compensation holograms onto the CFB facet enabling direct far field imaging in a simplified and robust manner. Robustness and field of view were increased by aperiodic and twisted fibers.

The use of a single multimode fiber (MMF) as a high-resolution endoscopic imaging tool is demonstrated. We show that the scrambled output of a MMF can be used for auto-fluorescence compressive imaging. By scanning a light spot across the proximal side of the fiber we can create uncorrelated speckle patterns at the output. Those patterns successively illuminate the biological sample and for each pattern the integrated intensity is recorded in epi-direction. An image of the brain tissue was computationally reconstructed using a regularization algorithm. Furthermore, the presented technique has potential in enhanced acquisition speed and in improving the resolution limit.

Adaptive optics and wavefront control enable advances in the transmission matrix measurement. However, despite its rich information content, the transmission matrix is usually only used to enhance light transmission through scattering media. For the first time, we present an assessment of the health status of retinal tissues using the transmission matrix measured by digital holography. Our data show that transmission matrix analysis can detect pathological changes in the retina and is promising for the development of label-free imaging biomarkers.