一种基于量子经典混合神经网络的激光散斑对比成像系统

A Laser Speckle Contrast Imaging System Based on a Quantum-Classical Hybrid Neural Network

摘要: 本研究旨在开发一种激光散斑对比成像(laser speckle contrast imaging, LSCI)系统，该系统可实现血流速度的定量测量。LSCI是一种非侵入性的光学成像技术，广泛应用于微循环血流速度监测，但现有LSCI系统无法实现血流速度定量测量，限制了其在临床中的广泛应用。本研究开发了一种基于量子经典混合神经网络的系统，该系统由硬件设计和算法设计两部分组成，主要实现数据采集、数据处理和血流速度预测功能。硬件设计部分以光照单元、采集单元和存储单元三部分为核心，算法设计部分通过混合模型实现血流速度的预测。活体验证中，系统在手指三个位置的血流速度曲线和传统算法的相关系数分别为0.924、0.867和0.899，同时可输出定量预测的局部血流速度曲线和血流动态变化图。

Abstract: The aim of this study is to develop a Laser Speckle Contrast Imaging (LSCI) system capable of quantitatively measuring blood flow velocity. LSCI is a non-invasive optical imaging technique widely used for monitoring microcirculatory blood flow velocity. However, current LSCI systems are unable to provide quantitative measurements of blood flow velocity, which limits their broader clinical application. This study introduces a system based on a quantum-classical hybrid neural network, comprising hardware and algorithm design, and is mainly responsible for data acquisition, processing, and blood flow velocity prediction. The hardware design centers around three core components: the illumination unit, acquisition unit, and storage unit. The algorithm design employs a hybrid model to predict blood flow velocity. In vivo validation showed that the correlation coefficients between the predicted blood flow velocity curves at three finger positions and those from traditional algorithms were 0.924, 0.867, and 0.899, respectively. The system is also capable of outputting quantitatively predicted local blood flow velocity curves and dynamic blood flow variation maps.