Publication detail

Segmentation of Atrial Activity in Intracardiac Electrograms (EGMs) Using Convolutional Neural Network (CNN) Trained on Small Imbalanced Dataset

HEJČ, J. POSPÍŠIL, D. NOVOTNÁ, P. PEŠL, M. JANOUŠEK, O. RONZHINA, M. STÁREK, Z.

Original Title

Segmentation of Atrial Activity in Intracardiac Electrograms (EGMs) Using Convolutional Neural Network (CNN) Trained on Small Imbalanced Dataset

Type

conference paper

Language

English

Original Abstract

Timing pattern of intracardiac atrial activity recorded by multipolar catheter in the coronary sinus (CS) provides insightful information about the type and approximate origin of common non-complex arrhythmias. Depending on the anatomy of the CS the atrial activity can be substantially disturbed by ventricular far field complex preventing accurate segmentation by conventional methods. In this paper, we present small clinically validated database of 326 surface and intracardiac electrocardiograms (ECG and IECG) and a simple deep learning framework for semantic beat-to-beat segmentation of atrial activity in CS recordings. The model is based on a residual convolutional neural network (CNN) combined with pyramidal upsampling label decoder. It is capable to recognize well between of atrial and ventricular signals recorded by decapolar CS catheter in multiple arrhytmic scenarios reaching dice score of 0.875 on evaluation dataset. To address a dataset size and imbalance issues, we have adopted several preprocessing and learning techniques with adequate evaluation of its impact on the model performance.

Keywords

intracardiac electrograms; atrial activity; convolutional neural network; imbalanced data; deep learning; arrhythmias

Authors

HEJČ, J.; POSPÍŠIL, D.; NOVOTNÁ, P.; PEŠL, M.; JANOUŠEK, O.; RONZHINA, M.; STÁREK, Z.

Released

18. 11. 2021

Publisher

Computing in Cardiology 2021

ISBN

0276-6574

Periodical

Computers in Cardiology

State

United States of America

Pages from

1

Pages to

4

Pages count

4

BibTex

@inproceedings{BUT174104,
  author="Jakub {Hejč} and David {Pospíšil} and Petra {Novotná} and Martin {Pešl} and Oto {Janoušek} and Marina {Filipenská} and Zdeněk {Stárek}",
  title="Segmentation of Atrial Activity in Intracardiac Electrograms (EGMs) Using Convolutional Neural Network (CNN) Trained on Small Imbalanced Dataset",
  booktitle="Computing in Cardiology 2021",
  year="2021",
  journal="Computers in Cardiology",
  pages="1--4",
  publisher="Computing in Cardiology 2021",
  issn="0276-6574"
}