Detail publikace

Automatic Detection of Strict Left Bundle Branch Block

Originální název

Automatic Detection of Strict Left Bundle Branch Block

Anglický název

Automatic Detection of Strict Left Bundle Branch Block

Jazyk

en

Originální abstrakt

Strict (true) left bundle branch block (tLBBB) ECG morphology is a new diagnostic marker in cardiology that was proposed to predict cardiac resynchronization therapy (CRT) responders. In this paper we present an algorithm for the automatic detection of tLBBB. This algorithm includes mid-QRS notching and slurring detection, QRS duration measurement and tLBBB morphology detection. All required morphologies are detected in the time domain using thresholding of simple features of signal. In order to test our algorithms, three experts labelled 78 ECG records (12 leads, fs = 5 kHz, 15 min); 51 records were labeled as tLBBB. The proposed algorithms were tested showing overall sensitivity and specificity 98 and 86%, respectively, in cases where all three experts reached full consensus (82% of the dataset). Our method showed lower sensitivity and higher specificity 96% and 88%, respectively, for the dataset including cases where experts mutually disagreed, consensus has been reached through expert discussion in these records.

Anglický abstrakt

Strict (true) left bundle branch block (tLBBB) ECG morphology is a new diagnostic marker in cardiology that was proposed to predict cardiac resynchronization therapy (CRT) responders. In this paper we present an algorithm for the automatic detection of tLBBB. This algorithm includes mid-QRS notching and slurring detection, QRS duration measurement and tLBBB morphology detection. All required morphologies are detected in the time domain using thresholding of simple features of signal. In order to test our algorithms, three experts labelled 78 ECG records (12 leads, fs = 5 kHz, 15 min); 51 records were labeled as tLBBB. The proposed algorithms were tested showing overall sensitivity and specificity 98 and 86%, respectively, in cases where all three experts reached full consensus (82% of the dataset). Our method showed lower sensitivity and higher specificity 96% and 88%, respectively, for the dataset including cases where experts mutually disagreed, consensus has been reached through expert discussion in these records.

BibTex


@inproceedings{BUT147400,
  author="Radovan {Smíšek} and Pavel {Jurák} and Ivo {Viščor} and Josef {Halámek} and Filip {Plešinger} and Magdaléna {Matejková} and Pavel {Leinveber} and Jana {Kolářová}",
  title="Automatic Detection of Strict Left Bundle Branch Block",
  annote="Strict (true) left bundle branch block (tLBBB) ECG morphology is a new diagnostic marker in cardiology that was proposed to predict cardiac resynchronization therapy (CRT) responders. In this paper we present an algorithm for the automatic detection of tLBBB. This algorithm includes mid-QRS notching and slurring detection, QRS duration measurement and tLBBB morphology detection. All required morphologies are detected in the time domain using thresholding of simple features of signal. In order to test our algorithms, three experts labelled 78 ECG records (12 leads, fs = 5 kHz, 15 min); 51 records were labeled as tLBBB. The proposed algorithms were tested showing overall sensitivity and specificity 98 and 86%, respectively, in cases where all three experts reached full consensus (82% of the dataset). Our method showed lower sensitivity and higher specificity 96% and 88%, respectively, for the dataset including cases where experts mutually disagreed, consensus has been reached through expert discussion in these records.",
  address="Springer Singapore",
  booktitle="World Congress on Medical Physics and Biomedical Engineering 2018",
  chapter="147400",
  doi="10.1007/978-981-10-9038-7_82",
  edition="IFMBE Proceedings",
  howpublished="online",
  institution="Springer Singapore",
  number="3",
  year="2018",
  month="may",
  pages="1--5",
  publisher="Springer Singapore",
  type="conference paper"
}