Naslov
Relevance of empirical mode decomposition for fetal heartbeat detection on smartphone devices

Autori
Vican, Ivan ; Kreković, Gordan ; Jambrošić, Kristian

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
The 8th Congress of the Alps Adria Acoustics Association – Conference Proceedings / Horvat, Marko ; Krhen, Miljenko - Zagreb : Hrvatsko akustičko društvo (HAD), 2018, 64-69

ISBN
978-953-95097-2-7

Skup
The 8th Congress of the Alps Adria Acoustics Association

Mjesto i datum
Zagreb, Hrvatska, 20.09.2018. - 21.09.2018

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
phonocardiography ; fetal heartbeat ; feature extraction ; feature ranking ; feature selection ; machine learning ; prenatal care

Sažetak
Fetal phonocardiography is a re-emerging method for extracting fetal heartbeat signals with a strong potential to be used as an easily accessible system in prenatal monitoring, especially if employed in conjunction with widespread electronic hardware. Since smartphone devices are going through rapid development of their processing power, sensory capabilities and network connectivity, they are becoming a powerful yet underutilized biomedical tool. Within this study we propose novel features for automatic fetal heartbeat detection based on intrinsic mode functions (IMF) gained through empirical mode decomposition. In order to show that more accurate detection can be achieved with IMF- based features added to the conventional set of audio features, we assessed feature relevance and usefulness using ranking and selection techniques. The results suggest that IMF-based features are relevant for the classification task and can improve prediction accuracy by 3.28%.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika

POVEZANOST RADA