Application of Independent Component Analysis to the Characterization of Atrial Fibrillation

• Main Authors: Jian-Hung Liu, 劉建宏
• Other Authors: Tsair Kao
• Format: Others
• Language: zh-TW
• Published: 2003
• Online Access: http://ndltd.ncl.edu.tw/handle/66485998060169920456

碩士 === 國立陽明大學 === 醫學工程研究所 === 91 === The widely accepted theory of mechanism of atrial fibrillation (AF) is that there will be four to six reentrant wavelets in the left and right atria. Recently, Independent component analysis (ICA) has been used to reveal hidden factors that underlie sets of random variables. In order to understand of the mechanism and characteristics of AF, we have used ICA technique to separate the possible independent reentrant wavelets in the atrium during AF. All animal experiments were performed with nine canines that weighed between 10 and 15 kg, the experiments were approved by the Taipei Veterans General Hospital, Taipei, Taiwan. Dogs were open chest and used plastic plank to record epicardial signals of right atrium. The recorded signals included sinus rhythm, during AF, after inject of antiarrhythmic drugs and before restoration to normal sinus rhythm. In ICA analysis, fixed-point algorithm was used to find minimization of mutual information. Principle component analysis (PCA) and coherence function determine the number of output components. In the sinus rhythm, only one signal component that has regular activation sequence was present. During AF, some wavelets that have higher frequency and irregular activation sequence were separated. After injected antiarrhythmic drug, there were still some reentrant wavelets. Compared with those during AF, both the number of wavelets and the frequencies of wavelets were decreased. Before restoration to normal sinus rhythm, wavelets reduced to less than two and least the rhythm of among wavelets can be re-controlled by the sinoatrial node. According to our results, we proposed a hypothesis that before the termination of AF, numbers of reentrant wavelets were reduced and component with frequency close to the upper limit of sinus rhythm. The location of this signal component was close to the area around SA node.