With the advancement of deep learning and neural networks, the computational demands for applications in wearable devices have grown exponentially. However, wearable devices also have strict requirements for long battery life, low power consumption, and compact size. In this work, we propose a scalable optoelectronic computing system based on an integrated optical convolution acceleration core. This system enables high-precision computation at the speed of light, achieving 7-bit accuracy while maintaining extremely low power consumption. It also demonstrates peak throughput of 3.2 TOPS (tera operations per second) in parallel processing. We have successfully demonstrated image convolution and the typical application of an interactive first-person perspective gesture recognition application based on depth information. The system achieves a comparable recognition accuracy to traditional electronic computation in all blind tests.

A critical function of flow cytometry is to count the concentration of blood cells, which helps in the diagnosis of certain diseases. However, the bulky nature of commercial flow cytometers makes such tests only available in hospitals or laboratories, hindering the spread of point-of-care testing (POCT), especially in underdeveloped areas. Here, we propose a smart Palm-size Optofluidic Hematology Analyzer based on a miniature fluorescence microscope and a microfluidic platform to lighten the device to improve its portability. This gadget has a dimension of 35 × 30 × 80 mm and a mass of 39 g, less than 5% of the weight of commercially available flow cytometers. Additionally, automatic leukocyte concentration detection has been realized through the integration of image processing and leukocyte counting algorithms. We compared the leukocyte concentration measurement between our approach and a hemocytometer using the Passing-Bablok analysis and achieved a correlation coefficient of 0.979. Through Bland-Altman analysis, we obtained the relationship between their differences and mean measurement values and established 95% limits of agreement, ranging from ?0.93×103 to 0.94×103 cells/μL. We anticipate that this device can be used widely for monitoring and treating diseases such as HIV and tumors beyond hospitals.

Mode-locked microcombs with flat spectral profiles provide the high signal-to-noise ratio and are in high demand for wavelength division multiplexing (WDM)-based applications, particularly in future high-capacity communication and parallel optical computing. Here, we present two solutions to generate local relatively flat spectral profiles. One microcavity with ultra-flat integrated dispersion is pumped to generate one relatively flat single soliton source spanning over 150 nm. Besides, one extraordinary soliton crystal with single vacancy demonstrates the local relatively flat microcomb lines when the inner soliton spacings are slightly irregular. Our work paves a new way for soliton-based applications owing to the relatively flat spectral characteristics.