高动态微伏级T波交替多域多通道融合检测研究

Microvolt T wave alternans (MTWA) under mobile monitoring environment displays high dynamic on temporal and space distribution, along with time-varying and non-Gaussian noise. As result, the detection of MTWA can only be achieved under resting and controllable conditions with limited clinical application in the risk prediction of the sudden cardiac death. In this project, the priori knowledge of signal and noise such as their rules of movement, change and development is deeply considered, therefore, a multi-domain and multi-channel fusion processing framework is proposed from the view of multiple feature domains of priori knowledge and multiple channels: First, the dynamic evolution law and multi-scale information of PQRST wave are comprehensively used to extract T wave in the time-scale domain and separate non Gaussian noise; Second, treating the ventricular repolarization as a quasi-sine wave oscillation signal, TWA space statistical domain fusion estimation method with circular frequency as a priori information is presented to solve the unsteady distribution based on carrier signal cycle stationary feature extraction theory. Finally, with the soft decision idea of Viterbi decoding, the TWA in spatial-evidence domain soft fusion decision method is presented to solve the problem of uneven spatial distribution, robustly detecting the high dynamic MTWA. The theoretical and technical foundation is laid to promote the MTWA detection from the controllable rest to moving portable environment.

移动监护环境下微伏级T波交替(MTWA)时空分布高动态、干扰噪声时变非高斯，致使当前MTWA检测只能在患者静息受控状态下实现，严重制约MTWA在心脏性猝死危险预测中的临床应用。本课题充分利用信号与干扰噪声的先验知识，分析研究其运动、变化、发展规律，从先验信息所在的多个特征空间与多个采集通道两方面入手，提出一种多域多通道融合分层处理方法：首先综合利用PQRST波动态演化规律与多尺度信息，从时间-尺度域融合提取T波，分离非高斯干扰噪声；其次把心室复极振荡看作一种拟正弦波振荡信号，从载波信号循环平稳特征提取理论出发，探索以循环频率为先验信息的TWA空间-统计域融合估计方法，解决时域分布非稳态问题；最后借鉴Viterbi译码软判决思想，探索空间-证据域TWA柔性融合判决方法，解决空域分布非均匀问题，实现高动态微伏级TWA鲁棒检测，为推进MTWA从静息受控检测到移动随身检测的跨越奠定理论与技术基础。

围绕微伏级TWA多通道融合检测与估计的相关理论与方法展开深入研究：针对TWA幅度弱、噪声干扰复杂、通道分布非均匀导致的通道间高冲突性信息综合利用困难的问题，提出了基于冲突信息重分配机制的微伏级TWA多通道融合检测方法，实现了复杂噪声干扰环境下微伏级TWA的可靠检测；在此基础上，进一步研究多通道TWA的高阶张量模型，提出了基于CP分解和重构的微伏级TWA多通道融合估计方法，实现了短时高动态TWA波形形态的有效估计。主要工作如下：.1、提出了一种基于信任函数的微伏级TWA的多通道融合检测方法。在多传感器决策级信息融合框架下，创新性地将信任函数理论应用于多通道冲突信息的描述，通过冲突信息重分配机制，解决高冲突性通道信息综合利用困难的问题。实验结果表明，在复杂噪声环境下，检测性能较现有算法平均提高2dB，检出率最高提升37个百分点。.2、提出了一种基于权重的微伏级TWA多通道融合检测方法。基于非等可靠信息融合思想，创新性地将证据权重与冲突信息重分配机制相结合，增强了非均匀分布TWA的检测能力。实验结果表明，综合检测性能(AUC)较现有算法平均提高了10个百分点，误检率小于5%的条件下，心肌梗塞患者的检出率提高了6个百分点。.3、提出了一种基于CP分解与重构的微伏级TWA多通道融合估计方法。在多传感器信号级信息融合框架下，创新性地将高阶张量应用于多通道TWA的描述，通过综合利用TWA在时域、空域和心拍域特性，增强短时高动态TWA波形估计能力。实验结果表明，在-15dB信噪比环境下，实现了16心拍短时TWA有效跟踪与估计。