Masked Training for Robust Arrhythmia Detection from Digitalized Multiple Layout ECG Images

2025-08-06

AI Summary

To address lead-wise temporal asynchrony and local signal occlusion arising from heterogeneous ECG image layouts across multi-hospital settings, this paper proposes PatchECG—a masked pretraining framework for adaptive representation learning. PatchECG models inter-lead collaborative dependencies via a masking mechanism, and integrates contrastive learning with a multi-head attention network to enable robust patch-level feature extraction and adaptive spatial focus. Trained jointly on PTB-XL and a custom asynchronous ECG dataset, the model achieves an average AUROC of 0.835. In external validation, it attains an AUROC of 0.778 for atrial fibrillation detection—improving to 0.893 under standardized 12×1 layout—surpassing ECGFounder by 0.111 and 0.190, respectively. This work is the first to introduce masked pretraining into multi-layout ECG analysis, significantly enhancing model generalizability and clinical applicability.

Technology Category

• Machine Learning: Multi-instance/Multi-view Learning
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Web Mining and Content Analysis: Large pretrained models with web data

Abstract

Electrocardiogram (ECG) as an important tool for diagnosing cardiovascular diseases such as arrhythmia. Due to the differences in ECG layouts used by different hospitals, the digitized signals exhibit asynchronous lead time and partial blackout loss, which poses a serious challenge to existing models. To address this challenge, the study introduced PatchECG, a framework for adaptive variable block count missing representation learning based on a masking training strategy, which automatically focuses on key patches with collaborative dependencies between leads, thereby achieving key recognition of arrhythmia in ECGs with different layouts. Experiments were conducted on the PTB-XL dataset and 21388 asynchronous ECG images generated using ECG image kit tool, using the 23 Subclasses as labels. The proposed method demonstrated strong robustness under different layouts, with average Area Under the Receiver Operating Characteristic Curve (AUROC) of 0.835 and remained stable (unchanged with layout changes). In external validation based on 400 real ECG images data from Chaoyang Hospital, the AUROC for atrial fibrillation diagnosis reached 0.778; On 12 x 1 layout ECGs, AUROC reaches 0.893. This result is superior to various classic interpolation and baseline methods, and compared to the current optimal large-scale pre-training model ECGFounder, it has improved by 0.111 and 0.19.

Problem

• Detect arrhythmia from diverse ECG layouts robustly
• Address asynchronous lead time and blackout loss issues
• Improve accuracy across variable hospital ECG formats

Innovation

• Masked training for robust ECG analysis
• Adaptive variable block missing learning
• Automatic key patch focus across layouts