Name
Title | ||
|---|---|---|
Classification Of Ecg Recordings With Neural Networks Based On Specific Morphological Features And Regularity Of The Signal |
Abstract | ||
|---|---|---|
The problem of detecting and analysing various kinds of pathologies in ECG signals has been profoundly researched in last years, but there is still no satisfying solution to distinguish such signals from normal or too noisy recordings.Our approach to solve this problem is based on analysis of ECG signals in time and frequency domain. It combines machine learning (neural networks, bagged trees) with standard methods of classification of cardiac arrhythmias and other pathologies, from which the features for the network are determined. We want to underline that all of the features used in classifying algorithm are based on signals morphology and other physiologically reasonable factors. We concentrate mostly on features resulting from comparison of QRS complex shapes and regularity of RR intervals.In presented approach, we decided not to feed the network with many random features, but to focus only on those adequate for ECG signal analysis (in other words, only those that can be easily understood by the physician). |
Year | DOI | Venue |
|---|---|---|
2017 | 10.22489/CinC.2017.356-350 | 2017 COMPUTING IN CARDIOLOGY (CINC) |
Field | DocType | Volume |
Frequency domain,Signal processing,Pattern recognition,Computer science,QRS complex,Artificial intelligence,Artificial neural network | Conference | 44 |
ISSN | Citations | PageRank |
2325-8861 | 0 | 0.34 |
References | Authors | |
2 | 2 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Katarzyna Stepien | 1 | 0 | 0.68 |
| iga grzegorczyk | 2 | 1 | 1.75 |