The rapid development of artificial intelligence and big data has increasingly heightened the demands on data center interconnection technologies. Terahertz wireless communication technology, which does not rely on physical cables, offers important interconnection form for data center. Our study introduces a new terahertz wireless interconnect system based on on-chip optical frequency comb for data center. It discusses the impact of waveguide second order dispersion on the generation of on-chip optical frequency combs and terahertz wireless interconnect performance in a data center environment. Simulation results show that using PAM4 encoding, the system based on an on-chip optical frequency comb can achieve a transmission rate of 120Gbps with bit error rate below 0.001 for data center terahertz wireless interconnects. This study lays a basic foundation for the future application of wireless terahertz interconnect in data center.

In this paper, we propose a multi-illuminant imaging system that captures the original smartphone camera responses under multiple LED light sources and reconstructs surface-spectral reflectance using locally weighted training samples. The method unfolds in three stages: (1) error calculation based on the captured camera response data to select locally optimal training samples for each test sample, (2) computation of a weighted coefficient matrix for the chosen locally optimal training samples, and (3) reconstructing surface-spectral reflectance employing wiener estimation, linear programming model, quadratic programming model, and different smartphone cameras. Experimental results demonstrate that the proposed methods significantly improve reconstruction accuracy, with the reconstructed goodness-of-fit coefficients consistently exceeding 0.9980, thereby confirming the reliability of the localized weighted model.

In dermatology, the mortality rate for melanoma is as high as 1.62%, imposing a significant burden on patients. Diagnosis primarily relies on the experience of physicians, who examine the skin’s color and texture. Traditional computer-assisted lesion segmentation has a low accuracy. However, with advancements in deep learning and the integration of various factors from medical imaging, the accuracy of lesion segmentation can be enhanced. Our skin disease detection method, based on deep learning, uses the GGAD-Net model and the SeDice loss function, with a weighting parameter of 0.5 for SeDice. On the ISIC 2018 dataset, after 50 epochs of training, the GGAD-Net model achieved an accuracy of 92.2%, a recall rate of 96%, a Dice score of 92.3%, and an IoU of 88.4%. The results prove that this model can effectively improve the performance of lesion segmentation, and in the future, we aim to further enhance accuracy and expand to other medical imaging segmentation applications.

The gradient in the helicity of optical field can exert the discriminatory force on the enantiomer, and this force is a concept well established within the dipole approximation for a Rayleigh particle. Here, with the aid of the multipole moment expansion theory, the formulae of the multipolar helicity-gradient force on a Mie chiral particle are derived. Moreover, such force induced by a tightly focused optical vortex is numerically simulated. The obtained results illustrate that the multipolar helicity-gradient force is not proportional to the gradient of optical helicity due to the nonlocal effect. The interception, recoil, electric, and magnetic components in such force are remarkably different. Besides, this force shows apparent dependence on the particles radius, beam topological charge and radio of the pupil radius to the beam waist radius. These findings can help us to understand the interplay of multipoles and may be exploited in chirality sensing, optomechanics, and manipulation.

Photoacoustic endoscopy (PE) is an emerging medical imaging technology that combines optical and acoustic imaging modalities, enabling high-resolution imaging of tissues inside the body. Compared to traditional piezoelectric sensors, optical sensors offer advantages in sensitivity, bandwidth, and size, making them suitable for PE applications. This paper demonstrates a high-order phase-shifted Bragg grating waveguide (HP-BGW) fabricated by two-photon polymerization 3D printing as an ultrasonic sensor. The grating is designed to achieve a high transmittance, and sharp line shape in resonant transmission so that enhancing the sensor's sensitivity. The proposed sensor is easy to manufacture and features a wide bandwidth and small size, making it suitable for photoacoustic endoscopy applications. The sensor was characterized using a spectrometer and ultrasound transducers, achieving a Q-factor of approximately 1×10^4 with a bandwidth of around 25MHz and a dynamic range of at least 29MHz.

Raman spectra are widely used in various fields, so it is crucial to realize the fast detection of Raman spectra. The removal of noise in Raman spectra is one of the difficulties in spectroscopic research, and this paper proposes a denoising and reconstruction method for spectral signals based on wavelet transform and weighted Wiener estimation. The wavelet transform is used to remove high-frequency noise, and then the Raman spectrum is reconstructed with the help of weighted Wiener estimation. The spectra in the range of 600-1800 are denoised and reconstructed using samples of body membranes and cells as research objects. The results show that the wavelet transform can effectively remove the noise from the spectra, while the weighted Wiener estimation can reconstruct the spectra accurately. Therefore, the spectral denoising reconstruction method based on wavelet transform and weighted Wiener estimation can improve the resolution of spectral reconstruction and show significant advantages in spectral reconstruction.

Due to the limitations of unknown light sources, object surface reflectivity, and imaging devices, it is still challenging to infer scene illumination based on the color representation of object surfaces. To further optimize the deep learning-based color constancy algorithm, we propose a two-channel weighted image color constancy network based on image patches. The network utilizes a feature fusion technique to combine shallow edge features and deep fine-grained features extracted from two network branches respectively. The strategy of using image patches increases the diversity of the dataset, and the weighting mechanism suppresses image noise to further improve the performance of the network. Experimental results on reprocessed Color Checker and NUS-8 datasets show that the network significantly improves performance while maintaining model complexity.

In this work, we describe a convolutional neural network (CNN) to accurately predict field lighting. In the network structure, feature learning and regression are integrated into an optimization process to form a more effective model for scene illumination estimation, and we have added an attention mechanism to reinforce learning. This approach is trained with ICVL data sets and tested with Foster HSI data for better performance. The stability of the proposed neural network for local illumination estimation and the improvement of the global illumination estimation performance are verified by experiments on the spatial illumination variation images.

The polarization effect significantly impacts the propagation of vector soliton in birefringent fibers, crucial for long-distance optical communication. With the development of artificial intelligence, deep learning methods, notably Physics-Informed Neural Network (PINN), have emerged as effective tools for solving Partial Differential Equations (PDE). In this work, we propose an Enhanced Multi-subnetwork PINN (EMPINN) with coupled double loss functions to predict vector soliton propagation. The advanced network accounts for predicting the propagation of vector soliton considered linear, elliptic, and circular polarization effects in two typical birefringent fibers. Across various cases, the predicted results exhibit low and reasonable errors compared to ground truth. The enhanced PINN method demonstrates stability, which reduces reliance on initial conditions, and improves computational efficiency. This approach offers a robust means to forecast ultrafast pulse propagation affected by polarization effects, advancing optical communication studies.

High-quality multimodal imaging technology is widely used in the biomedical field. Because the traditional transparent ultrasound transducer is opaque, the optical axis has to be another bypass, leading to the problems such as large volume, low integration and insufficient signal to noise ratio of the imaging equipment. Therefore, a novel optically transparent transducer is proposed herein for ultrasound transceiving. In this design, the acoustic-electric conversion element is made of a 100-μm-thickness lithium niobate crystal. Transparent indium tin oxide conductive layers are deposited onto the top and bottom surfaces of the circular or square crystal. A customized round metal housing is used for protection and encapsulation, and the optical resin acts as the acoustic matching layer and backing. Experimental results show that the new ultrasound transducer has a wider frequency band up to 40 MHz and a shorter response, compared to the commonly used piezoelectric ceramic ultrasound transducers on the market. The ultrasound transducer has the obvious advantages of compact and portable structure, wideband transparency, high sensitivity and high-frequency response. The proposed approach demonstrates the potential for seamless integration of ultrasound, photoacoustic, fluorescence, and other multimodal imaging techniques, thereby effectively advancing the development of photoacoustic imaging technology in biomedical applications.

This PDF file contains the front matter associated with SPIE Proceedings Volume 13227, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.