ECG Beat Detection and Pulmonary Dynamics Simulation

 

 


By:  Brent Goldman

Supervisor: Dr. Ilker Tunay

 

Washington University

Spring 2006

 
 
 
 
Abstract:
Catheters are commonly used by doctors to aid patients.  The purpose of this study is to filter out the movements caused by the body during catheterization.  The authors developed an algorithm that effectively locates the P wave and the R wave during a heart beat.  We also built a model using Matlab/Simulink that simulates the change in intrapleural pressure during a normal breathing cycle. By combining the ECG beat detection algorithm and the cardiovascular model, as well as, a circulation model, we can in effect eliminate the movements caused by the body.  When evaluated with the PhysioNet QT database, the ECG beat detection algorithm performed well against normal heart beats, and also against arrhythmias.  The results from the cardiovascular model are accurate with qualitative clinical descriptions of normal, quiet breathing. 
 
R Wave and P Wave Detection:
The features of an ECG are defined by the letters P, Q, R, S, and T.  Each letter corresponds to a wave or change in voltage.  The P wave signifies an atrial contraction.  By combining the Q, R, and S waves, we get a complex.  The QRS complex signifies the contraction of the left and right ventricles.  Finally, the T wave signifies repolarization of the ventricles.  In October of 2005, Quinghua Zhang published, “An Algorithm for Robust and Efficient Location of T-wave Ends in Electrocardiogram” [1].  The article describes in detail a method to detect the T wave.  In addition, it also shares the code used to detect both the T wave and the R wave.  According to the authors, the algorithm outperforms other algorithms in evaluating manually annotated ECG signals of the QT database available on the PhysioNet web site [1].  The first upward wave of the QRS complex is the R wave.  The R wave is the easiest to detect because it is the tallest wave.  Therefore, the algorithm used to detect the R wave is relatively simple compared to the other waves.  Once it is detected, there is an interval of time before the next QRS complex occurs.  This interval includes the T and P waves.  Once the T wave is detected, the only remaining wave is the P wave.  The figure below shows an ECG with the locations of the R wave, T-wave end, and P wave marked.  The R wave is marked with a solid blue line.  The T-wave end is marked with a red dashed line.  And the P wave is marked with a green dotted line.  The T wave in the figure below deflects downward, signifying an arrhythmia.         
 

Locations of the R, T-end, and P waves
 
 
 

 
Model of the Respiratory System:
A journal article published in 1995, “Anatomical and Physiological Simulation for Respiratory Mechanics,” derives equations necessary to develop a mathematical model of the respiratory system [2].  The model displays a lung as a single compartment.  Even though the system could be more complicated, the equations were shown to fit patient data well.  The figure below shows the main components of the respiratory system.    
 
 

Respiratory System
 
 
 
 

 
If we can measure the lung volume, we can determine the change in intrapleural pressure.  We tried to represent a normal, quiet breathing patient.  The average lung volume reaches 6 liters and follows a sinusoidal pattern.  We expressed the change in lung volume (vl) through |6*sin(.75*x)|.  Parameters that affect the intrapleural pressure (ppl) are airway resistance (Raw), lung compliance (Cl), and pressure in the airway opening (pao).  The values that we used, given in [2], are:
 

Normal, Quiet Breathing

R aw = 1.7 cmH20 per L/sec

Cl     = 200 mL per cm H20

pao = 0 cmH20

 
 
  
 
 
 
 
 
 
 

 

Plugging these values into the equation:  ppl = -vl / Cl  - R aw*v’l + pao  , we get figure(b) below.  Figure(a) represents the change in lung volume.  The results produced are accurate with qualitative clinical descriptions of normal, quiet breathing [2].    

 

(a)
 

(b)
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Results and Conclusion:

The P and R wave detection algorithms and the respiratory model performed well.  .  The algorithm had no erroneous detections of R waves. Our P wave detection algorithm has the following advantages.

 

·        filters out measurement noise

·        detects wave form morphological variations

·        calculation is easy

 
The respiratory model built from the equations derived in [3] was easy to implement.  The respiratory system is modeled as a series of deformable or rigid compartments that gas flows through [3].  Since we only modeled a single compartment, the accuracy of the model was affected, but not significantly enough to make a difference.  By combining the respiratory model and the ECG detection algorithms, we can filter out the movements caused by the body and effectively localize a catheter with respect to the heart.  We recommend implementing the programs developed in this paper based on their performance and significance in the medical field.        
 
 
 
 
 
Sources:
[1] Zhang, Qinghua, and Manriquez, Alfredo I. (Eds.). (2005). An Algorithm for Robust and Efficient Location of T-Wave Ends in Electrocardiogram. France: Instiut De Recherche en Informatique et Systemes Aleatoires.
 
[2] Kaye, Jonathon, and Primiano, Frank P. (Eds.). (1995). Anatomical and Physiological Simulation for Respiratory Mechanics.  J Image Guid Surg. 1995; 1(3):164-71.
<http://citeseer.ist.psu.edu/rd/25239409%2C169537%2C1%2C0.25%2CDownload/ftp%3AqSqqSqftp.cis.upenn.eduqSqpubqSqtraumaidqSqpapersqSqmrcas95.ps.gz>.