[1]刘文哲,郑一博,刘强,等.基于信号质量指数动态加权融合的多模态呼吸率估计方法[J].中国医学物理学杂志,2026,43(1):99-109.[doi:DOI:10.3969/j.issn.1005-202X.2026.01.013]
　LIU Wenzhe,ZHENG Yibo,LIU Qiang,et al.Signal quality index-based dynamic weighted fusion method for multimodal respiratory rate estimation[J].Chinese Journal of Medical Physics,2026,43(1):99-109.[doi:DOI:10.3969/j.issn.1005-202X.2026.01.013]
点击复制

基于信号质量指数动态加权融合的多模态呼吸率估计方法()

《中国医学物理学杂志》[ISSN:1005-202X/CN:44-1351/R]

卷:
43卷
期数:
2026年第1期
页码:
99-109
栏目:
医学信号处理与医学仪器
出版日期:
2026-01-26

文章信息/Info

Title:
Signal quality index-based dynamic weighted fusion method for multimodal respiratory rate estimation
文章编号:
1005-202X（2026）01-0099-11
作者:
刘文哲1; 郑一博2; 刘强2; 刘永伟1
1.河北地质大学信息工程学院， 河北 石家庄 050031； 2.河北省光电信息与地球探测技术重点实验室， 河北 石家庄 050031
Author(s):
LIU Wenzhe1;  ZHENG Yibo2;  LIU Qiang2;  LIU Yongwei1
1. School of Information Engineering, Hebei GEO University, Shijiazhuang 050031, China 2. Hebei Key Laboratory of Optoelectronic Information and Geo-detection Technology, Shijiazhuang 050031, China
关键词:
呼吸率估计; 信号质量指数; 心电图; 光电容积脉搏波; 多模态融合
Keywords:
Keywords: respiratory rate estimation signal quality index electrocardiogram photoplethysmogram multimodal fusion
分类号:
R318.04；TP274
DOI:
DOI:10.3969/j.issn.1005-202X.2026.01.013
文献标志码:
A
摘要:
提出一种基于信号质量指数（SQI）动态加权融合的多模态呼吸率估计方法，旨在提升非侵入式连续呼吸监测的精度与鲁棒性。该方法结合心电图衍生呼吸信号与光电容积脉搏波的希尔伯特包络信号，利用其生理互补性构建融合模型；通过实时计算SQI量化各模态信号的可靠性，并以此为据动态分配融合权重，从而自适应调整不同信号源的贡献比例，以应对信号质量波动。基于CapnoBase数据库的实验验证表明，本文方法在波形层面与参考呼吸信号保持高度一致，平均皮尔逊相关系数达0.818 1，波形重建误差显著降低；在呼吸率估计方面，本文方法表现出优异的准确性与稳定性，其平均绝对误差仅为0.36次/min，尤其在信号干扰场景下性能提升显著。本研究证实了基于SQI的动态多模态融合能有效增强呼吸估计系统的抗干扰能力和整体性能，为智慧医疗及可穿戴设备中的高可靠性连续呼吸监测提供创新且实用的解决方案。
Abstract:
Abstract: A dynamic weighted fusion method based on signal quality index (SQI) is proposed for multimodal respiratory rate estimation, aiming to improve the accuracy and robustness of non-invasive continuous respiratory monitoring. This method combines electrocardiogram-derived respiratory signals with the Hilbert envelope signals of photoplethysmogram, and constructs a fusion model by leveraging their physiological complementarity. Additionally, SQI is calculated in real time to quantify the reliability of each modal signal, and the quantified results are used to dynamically assign fusion weights, which enables adaptive adjustment of the contribution ratios of different signal sources, thereby effectively addressing signal quality fluctuations. Experimental validation on the CapnoBase database shows that the proposed SQI-based fusion method maintains a high degree of consistency with reference respiratory signal at the waveform level, achieving a mean Pearson correlation coefficient of 0.818 1 and a significant reduction in waveform reconstruction errors. In terms of respiratory rate estimation, the proposed method exhibits high accuracy and stability, with an average absolute error of only 0.36 breaths/min, and the performance improvement is particularly pronounced in signal interference scenarios. This study validates the effectiveness of SQI-based multimodal fusion in enhancing the anti-interference capability and overall performance of respiratory estimation systems, thus providing an innovative and practical solution for high-reliability continuous respiratory monitoring in smart healthcare and wearable devices.

备注/Memo

备注/Memo:
【收稿日期】2025-11-10 【基金项目】河北地质大学硕士在读研究生创新能力培养资助项目（CXZZDD202614） 【作者简介】刘文哲，硕士，研究方向：生物信号处理、智能医疗器械研发，E-mail: 923667201@qq.com 【通信作者】郑一博，博士，教授，研究方向：光纤传感、光学信号处理，E-mail: yibo_zheng@hgu.edu.cn
更新日期/Last Update: 2026-01-27