ADAPTIVE ESTIMATION OF THE QRS COMPLEX WAVE FEATURES OF THE ECG SIGNAL BY THE HERMIT

[deepusunny40]

ADAPTIVE ESTIMATION OF THE QRS COMPLEX WAVE FEATURES OF THE ECG SIGNAL BY THE HERMITE MODEL

[attachment=193]

INTRODUCTION

The Electrocardiographic signal ECG represents the electrical activity of the heart recorded on the
body surface
The analysis of this signal is the most common way to study and diagnose cardiac
dysfunctions
The ECG signal is characterized by its recurrent or periodic behavior with each beat
Each
recurrence is composed of a wave sequence P QRS complex and T waves where the most characteristic
wave set is the QRS complex
This complex represents the depolarization phenomenon of the ventricles
and therefore gives useful information about heart behavior
The beat to beat classi cation of the
QRS complexes will permit us to follow the heart evolution and to detect arrythmias like premature
ventricular contractions PVC
The ECG data compression allows e cient storage of a large amount
of ECG data and fast data transmission and signal processing for diagnosis
All these properties are of
great importance in health care units and to those that need data transmission to a central processing
unit

THE ADAPTIVE HERMITE MODEL ESTIMATION SYSTEM

In this section we present the Adaptive Hermite Model Estimation System AHMES to adaptively
calculate the cn and b coe cients
This system is based on the multiple input adaptive linear combiner
ALC with desired response Widrow and Stearns
The primary input to the AHMES is
the digitized QRS signal and the reference inputs are the digitized Hermite functions

the AHMES where there are two adaptation processes the weight adaptation and the parameter b
adaptation
The parameter b acts as an input to an Hermite function basis generator that produces the
elements of the reference input signals in the AHMES

EFFECTS OF MISALIGNMENT AT THE QRS DETECTION MARK

In this section we will analyze the e ect of a misalignment at the QRS mark location with respect to
the period of dk
The application of the AHMES needs to estimate the occurrence time of the repetitive
QRS signal Jan e et al b and in these cases errors can appear
When an error of
appears
in these estimations the reference inputs remain periodic with period L but the signal sk will change
its period to sk sk L
The value of
will be of a random nature and so will vary from period
to period


THE ADAPTIVE ALGORITHMS
The AHMES includes one adaptation process to get the estimation of the Hermite model coe cients
weight vector and another adaptation process to estimate the optimum width parameter b
We use
the least mean square LMS algorithm Widrow and Stearns to adjust the weight vector
that is implemented by the recursive expression