Identification And Rapid Recognition Of Spiral Tip Via Deterministic Learning Theory

Cardiovascular disease is the most common fatal disease not only in China but around the world as well,and the mortality rate of these kinds of diseases are increasing year by year.Most of patients who dead from sudden death suffers ventricular fibrillation(VF).The heart rate of VF patients is usually more than 300 beats per minute,so the mortality rate is very high.At present,the main treatment is to depolarize the heart with high voltage electricity to achieve the purpose of defibrillation.This kind of defibrillation is harmful to patients and has low energy utilization.Therefore,it is necessary to have a new understanding of ventricular fibrillation in medicine,so as to provide a new treatment idea for painless and low-energy defibrillation.From the perspective of excitable medium,it can be observed that the electrophysiological signal mode in healthy heart tissue is target wave,while that in the heart tissue of stall and ventricular fibrillation is spiral wave.Spiral wave is a kind of common spatiotemporal pattern system,and spiral tip is generally considered as the source of spiral wave.Through the analysis of the stability of spiral wave,it can be seen that the trajectory of spiral tip is directly related to the stability of spiral wave.The study of spiral tip system can observe infinite dimensional spiral wave system through two-dimensional system,intuitively obtain the information of system stability.It can help people understand ventricular fibrillation deeply and provide theoretical basis for treatment of VF.Deterministic learning theory is a kind of theory that uses RBF neural network to identify the dynamic system of periodic or regression trajectory locally.It uses this method to identify the unknown dynamic of the system.The dynamic information of the time-space system can be saved in the constant value neural network to realize the time invariant expression of the unknown dynamic of the time-varying system.Because the result of identification has been stored in the constant value neural network,using this kind of learned information,we can realize the dynamic fast pattern recognition of the system.In this paper,the dynamic identification and fast pattern recognition of spiral tips are studied.Firstly,the Barkley model of infinite dimensional spiral wave reactiondiffusion equation is finite-difference analyzed by five-point difference method,and the spiral tips are extracted from spiral waves by an iscontour method.In this paper,the common spiral tip dynamics are identified based on deterministic learning theory,and the identification results can be saved in a series of constant value neural networks.Based on the identified results,a recognition method is proposed to recognize spiral tip.Finally,a speed-frequency subsystem of spiral tip is proposed,which has the characteristics of good periodicity and short period.With this system,less neural network can be used for dynamic identification and pattern recognition of spiral tip.