High-fidelity multi-channel optical information transmission through scattering media.

High-fidelity optical information transmission through strongly scattering media is challenging, but is crucial for the applications such as the free-space optical communication in a haze or fog. Binarizing optical information can somehow suppress the disruptions caused by light scattering. However, this method gives a compromised communication throughput. Here, we propose high-fidelity multiplexing anti-scattering transmission (MAST). MAST encodes multiple bits into a complex-valued pattern, loads the complex-valued pattern to an optical field through modulation, and finally employs a scattering matrix-assisted retrieval technique to reconstruct the original information from the speckle patterns. In our demonstration, we multiplexed three channels and MAST achieved a high-fidelity transmission of 3072 (= 1024× 3) bits data per transmission and average transmission error as small as 0.06%.

Energy-efficient dispersion compensation for digital micromirror device.

Due to the wave nature of light, the diffraction pattern generated by an optical device is sensitive to the shift of wavelength. This fact significantly compromises the digital micromirror device (DMD) in applications, such as full-color holographic display and multi-color fluorescence microscopy. The existing dispersion compensation techniques for DMD involve adding diffractive elements, which causes a large amount of waste of optical energy. Here, we propose an energy-efficient dispersion compensation method, based on a dispersive prism, for DMD. This method simulates the diffraction pattern of the optical fields reflected from the DMD with an angular spectrum model. According to the simulation, a prism and a set of optical components are introduced to compensate for the angular dispersion of DMD-modulated optical fields. In the experiment, our method reduced the angular dispersion, between the 532 nm and 660 nm light beams, by a factor of ∼8.5.

Multi-view stitching phase measuring deflectometry for freeform specular surface metrology.

Phase measuring deflectometry (PMD) offers notable advantages for precision inspection of specular elements. Nevertheless, if confronts challenges when measuring freeform specular surfaces due to the dispersion of reflection rays from surfaces with high local slopes. Here, we propose a multi-view stitching PMD. It utilizes distinct sensors combining with a screen to capture the appearance of each region. After precisely calibrating the entire system to correct the absolute depth of each region, the appearances of all regions are precisely stitched together, reconstructing the comprehensive appearance of the surface. Through experimental setup, we measured the 3D morphology of a spherical lens with a curvature radius of 155.04 mm and a peak-to-valley (PV) value of 2.9 mm, which yielded a measurement accuracy of 5.3 µm (relative error: 0.18 %). Furthermore, we successfully measured the appearance of a curved mobile phone screen with local slopes ranging from -46.1° to 51.3°, and freeform acrylic sheet with local slopes ranging from -6.7° to 7.7° and a PV value of 5.3 mm.

High-frame-rate reconfigurable diffractive neural network based on superpixels.

The existing implementations of reconfigurable diffractive neural networks rely on both a liquid-crystal spatial light modulator and a digital micromirror device, which results in complexity in the alignment of the optical system and a constrained computational speed. Here, we propose a superpixel diffractive neural network that leverages solely a digital micromirror device to control the neuron bias and connection. This approach considerably simplifies the optical system and achieves a computational speed of 326 Hz per neural layer. We validate our method through experiments in digit classification, achieving an accuracy of 82.6%, and action recognition, attaining a perfect accuracy of 100%. Our findings demonstrate the effectiveness of the superpixel diffractive neural network in simplifying the optical system and enhancing computational speed, opening up new possibilities for real-time optical information processing applications.

Anti-scattering light focusing with full-polarization digital optical phase conjugation based on digital micromirror devices.

The high resolution of optical imaging and optogenetic stimulation in the deep tissue requires focusing light against strong scattering with high contrast. Digital optical phase conjugation (DOPC) has emerged recently as a promising solution for this requirement, because of its short latency. A digital micromirror device (DMD) in the implementation of DOPC enables a large number of modulation modes and a high speed of modulation both of which are important when dealing with a highly dynamic scattering medium. Here, we propose full-polarization DOPC (fpDOPC) in which two DMDs simultaneously modulate the two orthogonally polarized components of the optical field, respectively, to mitigate the effect of depolarization caused by strong scattering. We designed a simple system to overcome the difficulty of alignment encountered when modulating two polarized components independently. Our simulation and experiment showed that fpDOPC could generate a high-contrast focal spot, even though the polarization of light had been highly randomized by scattering. In comparison with the conventional method of modulating the polarization along a particular direction, fpDOPC can improve the peak to background ratio of the focal spot by a factor of two. This new technique has good potential in applications such as high-contrast light focusing in vivo.

Classification of breast cancer histology images using MSMV-PFENet.

Deep learning has been used extensively in histopathological image classification, but people in this field are still exploring new neural network architectures for more effective and efficient cancer diagnosis. Here, we propose multi-scale, multi-view progressive feature encoding network (MSMV-PFENet) for effective classification. With respect to the density of cell nuclei, we selected the regions potentially related to carcinogenesis at multiple scales from each view. The progressive feature encoding network then extracted the global and local features from these regions. A bidirectional long short-term memory analyzed the encoding vectors to get a category score, and finally the majority voting method integrated different views to classify the histopathological images. We tested our method on the breast cancer histology dataset from the ICIAR 2018 grand challenge. The proposed MSMV-PFENet achieved 93.0[Formula: see text] and 94.8[Formula: see text] accuracies at the patch and image levels, respectively. This method can potentially benefit the clinical cancer diagnosis.

Anti-scattering light focusing by fast wavefront shaping based on multi-pixel encoded digital-micromirror device.

Speed and enhancement are the two most important metrics for anti-scattering light focusing by wavefront shaping (WS), which requires a spatial light modulator with a large number of modulation modes and a fast speed of response. Among the commercial modulators, the digital-micromirror device (DMD) is the sole solution providing millions of modulation modes and a pattern rate higher than 20 kHz. Thus, it has the potential to accelerate the process of anti-scattering light focusing with a high enhancement. Nevertheless, modulating light in a binary mode by the DMD restricts both the speed and enhancement seriously. Here, we propose a multi-pixel encoded DMD-based WS method by combining multiple micromirrors into a single modulation unit to overcome the drawbacks of binary modulation. In addition, to efficiently optimize the wavefront, we adopted separable natural evolution strategies (SNES), which could carry out a global search against a noisy environment. Compared with the state-of-the-art DMD-based WS method, the proposed method increased the speed of optimization and enhancement of focus by a factor of 179 and 16, respectively. In our demonstration, we achieved 10 foci with homogeneous brightness at a high speed and formed W- and S-shape patterns against the scattering medium. The experimental results suggest that the proposed method will pave a new avenue for WS in the applications of biomedical imaging, photon therapy, optogenetics, dynamic holographic display, etc.