A new approach for identification of heartbeats in multimodal physiological signals
| Autor: | Ramesh Kumar Sunkaria, Omkar Singh |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Computer science
Multimodal data 0206 medical engineering Biomedical Engineering Data fragment Cardiac activity 02 engineering and technology Signal 03 medical and health sciences Electrocardiography 0302 clinical medicine Signal quality Heart Rate Humans Arterial Pressure Noise level Training set business.industry Pattern recognition Heart Signal Processing Computer-Assisted General Medicine 020601 biomedical engineering Identification (information) cardiovascular system Artificial intelligence business 030217 neurology & neurosurgery Algorithms circulatory and respiratory physiology |
| Zdroj: | Journal of medical engineeringtechnology. 42(3) |
| ISSN: | 1464-522X |
| Popis: | In this paper, a technique is proposed for detection of heartbeats in multimodal data. Recording of multiple physiological signals from the same subject is common practice nowadays. Multiple physiological signals are generally available but are processed separately without taking into consideration information from other relevant signals. The heartbeats are generally detected from R peaks in electrocardiogram (ECG) signal, however, if ECG is noisy, other signals reflecting the cardiac activity may be used for identifying heartbeats. This paper describes a new method for detection of heartbeats using ECG and arterial blood pressure (ABP) signals. The physiological data are segmented into various fragments and signal quality is determined using the judgment of noise level. If the ECG data fragment is noisy, heartbeats are computed from the ABP fragment. The evaluation was performed on training data set of computing in cardiology challenge 2014. The proposed methodology has resulted in better detection accuracy as compared to the unimodal methods. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
| Nepřihlášeným uživatelům se plný text nezobrazuje | K zobrazení výsledku je třeba se přihlásit. |