Title :
Using stochastic complexity for ECG signal analysis
Author :
Aldea, Rafael ; Giurcaneanu, Ciprian Doru
Author_Institution :
Inst. of Signal Process., Tampere Univ. of Technol., Tampere, Finland
Abstract :
Applying the MDL based digital signal segmentation method proposed in [1], a new QRS detection algorithm is designed and tested. It is experimentally shown that the algorithm performs better than other 11 algorithms, in terms of False Positive/False Negative estimations. The newly defined Weighted Diagnostic Distortion (WDD) measure[2] is computed for annotated files from MIT-BIH database to evaluate the accuracy of the detection algorithm and, also, to verify how well the P,T-waves are conserved by the broken line approximation.
Keywords :
electrocardiography; estimation theory; medical signal detection; stochastic processes; ECG signal analysis; MDL based digital signal segmentation method; MIT-BIH database; P,T-waves; QRS detection algorithm; WDD measure; false negative estimations; false positive estimations; signal detection algorithm; stochastic complexity; weighted diagnostic distortion measure; Abstracts; Approximation algorithms; Approximation methods; Indexes; Yttrium;
Conference_Titel :
Signal Processing Conference, 2002 11th European
Conference_Location :
Toulouse