Two Will Do: CNN With Asymmetric Loss, Self-Learning Label Correction, and Hand-Crafted Features for Imbalanced Multi-Label ECG Data Classification

In this work, we present a machine learning approach that is able to classify 30 cardiac abnormalities from an arbitrary number of electrocardiogram (ECG) leads. Features extracted by a deep convolutional neural network are combined with hand-crafted features (demographic, morphological, and heart r...

Full description

Saved in:
Bibliographic Details
Published in2021 Computing in Cardiology (CinC) Vol. 48; pp. 1 - 4
Main Authors Vazquez, Cristina Gallego, Breuss, Alexander, Gnarra, Oriella, Portmann, Julian, Da Poian, Giulia
Format Conference Proceeding
LanguageEnglish
Published Creative Commons 13.09.2021
Subjects
Electrocardiography
Feature extraction
Machine learning
Measurement
Multilayer perceptrons
Robustness
Training
Online AccessGet full text

Cover

More Information
Summary:In this work, we present a machine learning approach that is able to classify 30 cardiac abnormalities from an arbitrary number of electrocardiogram (ECG) leads. Features extracted by a deep convolutional neural network are combined with hand-crafted features (demographic, morphological, and heart rate variability metrics) and fed into a multilayer perceptron. We employ an Asymmetric Loss (ASL) function, which enables the model to focus on hard but under-represented samples. To mitigate the issue of ground-truth mislabeling and to provide robustness, we investigate the use of a self-learning label correction method that iteratively estimates correct labels during training. Our team SMS+1 placed 7 th on the unseen test set, with an overall challenge score of 0.51, and 0.52, 0.45, 0.50, 0.50, and 0.49 for twelve-, six-, four-, three-, and two-lead, respectively. Our model maintains similar diagnostic potential on both standard twelve-lead ECGs and reduced-lead ECGs.
ISSN:2325-887X
DOI:10.23919/CinC53138.2021.9662741