ECGXtract: Deep Learning-based ECG Feature Extraction for Automated CVD Diagnosis

TL;DR

ECGXtract introduces a CNN-based, interpretable ECG feature extractor that directly predicts temporal and morphological descriptors aligned with clinically validated ground-truth features. The study demonstrates strong correlations with ground truth across global and lead-specific features, with Lead II often providing the best performance and 100 Hz sampling preserving accuracy while reducing compute. Semantic grouping improves efficiency for global features, whereas lead-specific features benefit from specialized, smaller-group models. Compared to ECGdeli, ECGXtract achieves higher correspondence to ground truth in most features, highlighting its potential for scalable, interpretable ECG analysis in limited-resource settings and real-time deployments.

Abstract

This paper presents ECGXtract, a deep learning-based approach for interpretable ECG feature extraction, addressing the limitations of traditional signal processing and black-box machine learning methods. In particular, we develop convolutional neural network models capable of extracting both temporal and morphological features with strong correlations to a clinically validated ground truth. Initially, each model is trained to extract a single feature, ensuring precise and interpretable outputs. A series of experiments is then carried out to evaluate the proposed method across multiple setups, including global versus lead-specific features, different sampling frequencies, and comparisons with other approaches such as ECGdeli. Our findings show that ECGXtract achieves robust performance across most features with a mean correlation score of 0.80 with the ground truth for global features, with lead II consistently providing the best results. For lead-specific features, ECGXtract achieves a mean correlation score of 0.822. Moreover, ECGXtract achieves superior results to the state-of-the-art open source ECGdeli as it got a higher correlation score with the ground truth in 90% of the features. Furthermore, we explore the feasibility of extracting multiple features simultaneously utilizing a single model. Semantic grouping is proved to be effective for global features, while large-scale grouping and lead-specific multi-output models show notable performance drops. These results highlight the potential of structured grouping strategies to balance the computational efficiency vs. model accuracy, paving the way for more scalable and clinically interpretable ECG feature extraction systems in limited resource settings.

Figures (6)

• Figure 1: A sample ECG waveform.
• Figure 2: Main stages in ECGXtract.
• Figure 3: The proposed CNN model architecture for ECG feature extraction.
• Figure 4: PCC performance across different grouping configurations for lead-specific features.
• Figure 5: Correlations of model trained on 500 Hz signals and models trained on 100 Hz signals.