光谱学与光谱分析, 2020, 40 (7): 2048, 网络出版: 2020-12-04  

扩展的卡尔曼滤波在近红外光谱提取脑血流信号中的研究

Research on Extended Kalman Filter in Extracting Cerebral Blood Flow Signals by Near Infrared Spectroscopy
作者单位
1 吉林大学, 吉林 长春 130061
2 吉林大学珠海学院, 广东 珠海 519041
3 吉林大学第一医院, 吉林 长春 130061
摘要
脑血流中的血红蛋白有两种: 氧合血红蛋白(HbO2)和还原血红蛋白(HbR)。 这两种血红蛋白在脑血流中浓度的变化可以反应脑部神经活动, 提取其浓度变化信号可以为如癫痫病灶定位、 抑郁等相关疾病的诊断和治疗提供依据和参考。 目前, 使用近红外光谱提取脑血流信号的算法有EEMD-ICA法主成分分析法(PCA)、 独立成分分析法(ICA)、 相干平均法、 自适应滤波等, 这些算法在对近红外脑神经活动信号提取时都有各自的特点和优势, 但都重视如呼吸、 眼动等各种生理干扰, 忽视了测量过程中符合高斯分布的测量干扰, 如仪器精密度、 信号传输中的串扰等。 为了提取脑血流中氧合血红蛋白(HbO2)和还原血红蛋白(HbR)浓度变化信号, 设计了功能性近红外光谱(fNIRS)的脑血流参数采集装置, 选择波长为750和830 nm的二极管近红外光源采集脑部血流变化信号, 采用扩展的卡尔曼滤波(EKF)算法, 把生理干扰和测量干扰建立对应的数学模型, 使用基于误差平方和最小的原理进行递归计算, 通过对下一时刻系统的初步状态估计以及测量得出的反馈相结合, 得到该时刻无限逼近真实值的状态估计, 结合修正的朗伯比尔定律(Lambert-Beer law), 将光密度信号的变化转换为氧合血红蛋白(HbO2)和还原血红蛋白(HbR)浓度变化信号。 结果表明: 所提方法可以有效去除符合高斯分布的测量干扰, 在Valsava实验和视觉诱发实验中, 可以提取出脑血流中氧合血红蛋白(HbO2)和还原血红蛋白(HbR)浓度变化曲线, 和主流的EEMD提取脑信号算法比对其RMSE值提高了0.96%, r值提高了0.6%, 表明提出的方法有一定的优越性。 所提方法为相关脑部疾病诊断等提供了有效的脑神经活动探测方法。
Abstract
There are two types of hemoglobin in the cerebral blood stream: oxygenated hemoglobin (HbO2) and reduced hemoglobin (HbR). The changes in the concentration of these two hemoglobins in the cerebral blood flow can reflect the neural activity in the brain. Extracting the signals of concentration changes can provide basis and reference for the diagnosis and treatment of related diseases such as epilepsy focus localization and depression. At present, algorithms for extracting cerebral blood flow signals using near-infrared spectroscopy include the EEMD-ICA method principal component analysis (PCA) , independent component analysis (ICA), the coherent averaging method, Adaptive filtering, etc. The above algorithms have their own characteristics and advantages in the extraction of near-infrared brain neural activity signals. However, the above methods all pay attention to various physiological interferences such as respiration and eye movement and ignore measurement interferences that conform to Gaussian distribution during measurements, such as instrument precision and crosstalk in signal transmission. In order to extract signals of changes in the concentration of oxygenated hemoglobin (HbO2) and reduced hemoglobin (HbR) in cerebral blood flow, a functional near infrared spectroscopy (fNIRS) cerebral blood flow parameter acq uisition device is designed in this article. In the device, a light source Diode near-infrared light sources with wavelengths of 750 and 830 nm were selected to collect brain blood flow changes. The extended Klaman Filter (EKF) algorithm was used to establish a corresponding mathematical model of physiological interference and measurement interference. Perform recursive calculation with the minimum principle, and combine the initial state estimation of the system at the next moment with the measured feedback to obtain a state estimate of infinitely close to the real value at that moment.), The change of the optical density signal is converted into a signal of change in oxygenated hemoglobin (HbO2) and reduced hemoglobin (HbR) concentration. The results show that the method proposed in this paper can effectively remove the measurement interference that conforms to the Gaussian distribution. In the Valsava experiment and the visual evoked experiment, the curve of changes in the concentration of oxygenated hemoglobin (HbO2) and reduced hemoglobin (HbR) in the cerebral blood flow can be extracted. Compared with the mainstream EEMD algorithm for extracting brain signals, its RMSE value is increased by 0.96%, and r value is increased by 0.6%, which indicates that the proposed method has certain advantages. The method proposed in this paper provides an effective method for detecting neural activity in related brain diseases.

刘颂阳, 刘光达, 刘卓娅, 邱吉庆, 蔡靖, 朱展鹏, 张程, 齐远, 张尚. 扩展的卡尔曼滤波在近红外光谱提取脑血流信号中的研究[J]. 光谱学与光谱分析, 2020, 40(7): 2048. LIU Song-yang, LIU Guang-da, LIU Zhuo-ya, QIU Ji-qing, CAI Jing, ZHU Zhan-peng, ZHAGN Cheng, QI Yuan, ZHANG Shang. Research on Extended Kalman Filter in Extracting Cerebral Blood Flow Signals by Near Infrared Spectroscopy[J]. Spectroscopy and Spectral Analysis, 2020, 40(7): 2048.

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!