Reconstruction of 12-Lead Electrocardiogram from a Three-Lead Patch-Type Device Using a LSTM Network

Area of Science:

• Medical Instrumentation
• Biomedical Engineering
• Signal Processing

Background:

• Standard 12-lead electrocardiogram (ECG) reconstruction from patch-type devices remains a significant challenge.
• Previous attempts using linear regression algorithms have been unsatisfactory due to limitations in capturing signal components.
• Patch-type devices often struggle with reduced horizontal vector components from anterior chest placement.

Purpose of the Study:

• To design hardware for a three-lead patch-type ECG device.
• To develop and implement a long short-term memory (LSTM) network for reconstructing a standard 12-lead ECG.
• To evaluate the feasibility and clinical equivalence of the reconstructed ECG compared to a standard 12-lead ECG.

Main Methods:

• Hardware design for a three-lead patch-type ECG.
• Application of a long short-term memory (LSTM) network for ECG signal reconstruction.
• Comparative analysis of reconstructed 12-lead ECGs against standard 12-lead ECGs in healthy subjects and patients with pathologies.

Main Results:

• The LSTM network demonstrated superior performance over linear regression in reconstructing ECG signals.
• The average correlation coefficient between the reconstructed and standard 12-lead ECGs was 0.95.
• The reconstructed ECG showed identical ability to detect pathologic abnormalities as the standard ECG.

Conclusions:

• Reconstruction of a standard 12-lead ECG from a three-lead patch-type device is feasible.
• The LSTM network-based reconstruction offers a viable and equivalent alternative to conventional 12-lead ECGs.
• This technology has the potential to improve ambulatory cardiac monitoring and diagnostics.