Advanced P Wave Detection in Ecg Signals During Pathology: Evaluation in Different Arrhythmia Contexts
| Autor: | Radovan Smisek, Lucie Marsanova, Andrea Němcová, Martin Vitek, Lukas Smital |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2019 |
| Předmět: |
Pathology
medicine.medical_specialty Databases Factual Computer science 0206 medical engineering lcsh:Medicine 02 engineering and technology 030204 cardiovascular system & hematology Article 03 medical and health sciences Electrocardiography 0302 clinical medicine medicine Humans lcsh:Science Cardiac device therapy Multidisciplinary lcsh:R P wave Phasor Arrhythmias Cardiac Signal Processing Computer-Assisted 020601 biomedical engineering lcsh:Q Ecg signal Biomedical engineering Algorithms |
| Zdroj: | Scientific Reports Scientific Reports, Vol 9, Iss 1, Pp 1-11 (2019) |
| ISSN: | 2045-2322 |
| Popis: | Reliable P wave detection is necessary for accurate and automatic electrocardiogram (ECG) analysis. Currently, methods for P wave detection in physiological conditions are well-described and efficient. However, methods for P wave detection during pathology are not generally found in the literature, or their performance is insufficient. This work introduces a novel method, based on a phasor transform, as well as innovative rules that improve P wave detection during pathology. These rules are based on the extraction of a heartbeats’ morphological features and knowledge of heart manifestation during both physiological and pathological conditions. To properly evaluate the performance of the proposed algorithm in pathological conditions, a standard database with a sufficient number of reference P wave positions is needed. However, such a database did not exist. Thus, ECG experts annotated 12 chosen pathological records from the MIT-BIH Arrhythmia Database. These annotations are publicly available via Physionet. The algorithm performance was also validated using physiological records from the MIT-BIH Arrhythmia and QT databases. The results for physiological signals were Se = 98.42% and PP = 99.98%, which is comparable to other methods. For pathological signals, the proposed method reached Se = 96.40% and PP = 85.84%, which greatly outperforms other methods. This improvement represents a huge step towards fully automated analysis systems being respected by ECG experts. These systems are necessary, particularly in the area of long-term monitoring. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|