Adaptive Hermite models for ECG data compression: performance and evaluation with automatic wave detection

An orthogonal transformation based on Hermite functions is proposed as a method for ECG data compression. In order to apply the procedure four signal windows are selected in each beat, corresponding to the principal ECG features: P wave, QRS complex, ST segment and T wave. The performance of the method is analysed calculating the compression ratio (CR) and the relative mean-square error (MSE) in each window and in the whole beat. The method has been applied to ECG records from MIT/BIH arrhythmia database. In normal beats with a CR=11.6, the authors have obtained a MSE=(0.09/spl plusmn/0.02)%. In ECG signals containing normal beats and multiform PVCs a MSE=(0.56/spl plusmn/3.41)% is obtained, with a CR=10.3. To analyse the clinical applicability of the method, the algorithm was evaluated with an automatic wave detection program. Differences between the automatic measures in the original signal and in the reconstructed signal were compared and shown a good agreement.<<ETX>>

[1]  P. Caminal,et al.  Automatic wave onset and offset determination in ECG signals: Validation with the CSE database , 1992, Proceedings Computers in Cardiology.

[2]  W. A. Coberly,et al.  ECG data compression techniques-a unified approach , 1990, IEEE Transactions on Biomedical Engineering.

[3]  Olle Pahlm,et al.  A Method for Evaluation of QRS Shape Features Using a Mathematical Model for the ECG , 1981, IEEE Transactions on Biomedical Engineering.

[4]  P. Caminal,et al.  Adaptive feature extraction for QRS classification and ectopic beat detection , 1991, [1991] Proceedings Computers in Cardiology.

[5]  Gene H. Hostetter,et al.  Scan-Along Polygonal Approximation for Data Compression of Electrocardiograms , 1983, IEEE Transactions on Biomedical Engineering.