摘要
减少导管射频消融术中心室异位起搏点的定位时间,可减少射线暴露时间并降低导管术中血栓发生的风险。采用基于虚拟人所构建的高精度的全心脏模型,探讨以多次正向仿真以求解逆问题的方法,对异位起搏点的定位问题进行仿真研究。以心内膜某片区域为例,设定不同位置异位起搏点,研究其位置和异常的体表电位标测图(BSPM)之QRS等积分图之间的关系;构建神经网络并进行泛化能力和人为加噪情况下的稳定性测试。神经网络对异位起搏点分区初步定位的精度达10 mm×5 mm,准确率为25/31(不加噪声)和23/31(信噪比2 dB)。在此基础上,提出使用神经网络算法对异位起搏点初步定位,并通过BSPM误差相似度的定量分析,指导消融术中导管移动方向进行更加精细定位的方案。所提出的方法可有效克服单纯数学求解逆问题中的病态特性,同时显著减小了计算量,为进一步的深入研究提供了有益的思路和研究基础。
In radiofrequency catheter ablation （RFCA）, it is crucial to locate the proarrhythmia ectopic pacemaker before the ablation procedure. Research has been focused on locating the ventricular ectopic pacemakers non- invasively, precisely and rapidly. We proposed to explore the ECG inverse problem on the base of multiple simulation of body surface potential mapping （ BSPM）, for the purpose of locating ventricular ectopic pacemakers. Using our constructed high resolution whole heart model, ventricular arrhythmic BSPM was simulated. Different ectopic pacemakers were set on a certain region of endocardium, and the relation between paced location and QRS integral maps of paced BSPM was investigated. An artificial neural network （ANN） was designed and trained to localize paced points at a partition density of 10 mm x 5 ram, and its generalization capability was tested using simulated test sets with and without noise. The overall accuracy was 25/31 without noise and 23/31 with signal-to-noise ratio of 2dB. At last, a method, employing noninvasive BSPM, was proposed to facilitate endocardiac mapping and accelerate ectopic pacemaker localization according to clinical catheter paced mapping procedure. ANN was used to localize the ectopic point to a small region, then the catheter could move towards the direction derived by analyzing the quantitative relation between the abnormal pacemaker point location and BSPM. The proposed method has the potential to overcome the ill-posed character in solving ECG inverse problem using mathematical theories solely, and decrease the computational load as well.
出处
《中国生物医学工程学报》
CAS
CSCD
北大核心
2009年第3期377-385,共9页
Chinese Journal of Biomedical Engineering
关键词
全心模型
心电逆问题
导管射频消融术
神经网络
心室异位起搏点
whole heart model
the ECG inverse problem
radiofrequency catheter ablation （RFCA）
artificial neural network
ventricular ectopic pacemaker
