Detection of cardiovascular abnormalities using peak detection and adaptive thresholding: A synthetic and real time approach

In this paper a new method is proposed based on “modified thresholding algorithm” for diagnosing the Heart Diseases. Gaussian Kernel where used for synthesis of artificial ECG for testing the algorithm. A three dimensional dynamic model based on the single dipole model of the heart together with a realistic ECG noise model is used. Different noise sources like white noise, colored noise, real muscle artifacts, real electrode movements, real baseline wander, mixture of real baseline wander, electrode movements, and muscle artifacts are analysed and filtered. Real Time ECG data files of various patients of different age, sex, disease is tested. Using a low sensitivity analog filters ECG real time reading also been recorded and tested. Based on the information of the identified QRS complexes, the P waves and the T waves are detected. ECG classification is then carried out using the RR interval duration. The classification algorithm is trained to recognize four types of beat and will be used to find the cardiovascular abnormalities. Most automatic ECG diagnosis techniques require an accurate detection of the QRS complexes. So to maintain accuracy the tested results is been compared with the annotations.