Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Reconstructing 12-Lead ECG from 3-Lead ECG using Variational Autoencoder to Improve Cardiac Disease Detection of Wearable ECG Devices

Xinyan Guan, Yongfan Lai, Jiarui Jin, Jun Li, Haoyu Wang, Qinghao Zhao, Deyun Zhang, Shijia Geng, Shenda Hong

TL;DR

WearECG, a Variational Autoencoder (VAE) method that reconstructs twelve-lead ECGs from three leads: II, V1, and V5, demonstrates the potential of generative modeling for ECG reconstruction and its implications for scalable, low-cost cardiac screening.

Abstract

Twelve-lead electrocardiograms (ECGs) are the clinical gold standard for cardiac diagnosis, providing comprehensive spatial coverage of the heart necessary to detect conditions such as myocardial infarction (MI). However, their lack of portability limits continuous and large-scale use. Three-lead ECG systems are widely used in wearable devices due to their simplicity and mobility, but they often fail to capture pathologies in unmeasured regions. To address this, we propose WearECG, a Variational Autoencoder (VAE) method that reconstructs twelve-lead ECGs from three leads: II, V1, and V5. Our model includes architectural improvements to better capture temporal and spatial dependencies in ECG signals. We evaluate generation quality using MSE, MAE, and Frechet Inception Distance (FID), and assess clinical validity via a Turing test with expert cardiologists. To further validate diagnostic utility, we fine-tune ECGFounder, a large-scale pretrained ECG model, on a multi-label classification task involving over 40 cardiac conditions, including six different myocardial infarction locations, using both real and generated signals. Experiments on the MIMIC dataset show that our method produces physiologically realistic and diagnostically informative signals, with robust performance in downstream tasks. This work demonstrates the potential of generative modeling for ECG reconstruction and its implications for scalable, low-cost cardiac screening.

Reconstructing 12-Lead ECG from 3-Lead ECG using Variational Autoencoder to Improve Cardiac Disease Detection of Wearable ECG Devices

TL;DR

WearECG, a Variational Autoencoder (VAE) method that reconstructs twelve-lead ECGs from three leads: II, V1, and V5, demonstrates the potential of generative modeling for ECG reconstruction and its implications for scalable, low-cost cardiac screening.

Abstract

Twelve-lead electrocardiograms (ECGs) are the clinical gold standard for cardiac diagnosis, providing comprehensive spatial coverage of the heart necessary to detect conditions such as myocardial infarction (MI). However, their lack of portability limits continuous and large-scale use. Three-lead ECG systems are widely used in wearable devices due to their simplicity and mobility, but they often fail to capture pathologies in unmeasured regions. To address this, we propose WearECG, a Variational Autoencoder (VAE) method that reconstructs twelve-lead ECGs from three leads: II, V1, and V5. Our model includes architectural improvements to better capture temporal and spatial dependencies in ECG signals. We evaluate generation quality using MSE, MAE, and Frechet Inception Distance (FID), and assess clinical validity via a Turing test with expert cardiologists. To further validate diagnostic utility, we fine-tune ECGFounder, a large-scale pretrained ECG model, on a multi-label classification task involving over 40 cardiac conditions, including six different myocardial infarction locations, using both real and generated signals. Experiments on the MIMIC dataset show that our method produces physiologically realistic and diagnostically informative signals, with robust performance in downstream tasks. This work demonstrates the potential of generative modeling for ECG reconstruction and its implications for scalable, low-cost cardiac screening.

Paper Structure

This paper contains 22 sections, 13 equations, 5 figures, 4 tables.

Table of Contents

  1. ABSTRACT
  2. KEYWORDS
  3. INTRODUCTION
  4. RESULTS
  5. DISCUSSION
  6. METHODS
  7. RESOURCE AVAILABILITY
  8. ACKNOWLEDGMENTS
  9. DECLARATION OF INTERESTS

Figures (5)

  • Figure 1: Overview of our framework. (A) Three-lead ECG signals are fed into a generative model to reconstruct full 12-lead ECGs. (B) The reconstructed ECGs are evaluated via multiple strategies, including signal-level metrics, fine-tuned disease classification using ECGFounder and cardiologist-involved Turing test.
  • Figure 2: Spatial relationship among leads II V1 V5
  • Figure 3: Confusion matrix for model-assisted myocardial infarction (MI) detection.
  • Figure 4: Framework Overview
  • Figure 5: Representative examples of reconstructed 12-lead ECGs from the II/V1/V5 input configuration. Reconstructions closely match the ground truth morphology, including P-wave, QRS complex, and T-wave patterns.