Key result
Hybrid U-Net and ResNet model classifies ECG heartbeats with ~100% accuracy.
Why the study?
State-of-the-art ECG analysis algorithms rely on digital signal processing techniques that require considerable effort to re-tune rules for new morphologies, prompting development of a three-stage hierarchical deep learning framework for segmentation and beat recognition.
Does a three-stage hierarchical deep learning framework using U-Net and ResNet accurately segment ECG waveforms and classify cardiac beats in ambulatory ECG recordings from the PhysioNet QT Database?
Does a three-stage hierarchical deep learning framework using U-Net and ResNet accurately segment ECG waveforms and classify cardiac beats in ambulatory ECG recordings from the PhysioNet QT Database?
A three-stage hierarchical deep learning framework combining attention-guided U-Net segmentation and a 1D-ResNet classifier achieves near-saturated waveform localization (Stage 2 Macro F1 0.9992) and high accuracy (99.70%) for four-class ECG beat recognition on the PhysioNet QT Database.
Electrocardiogram (ECG) analysis is a key step in many clinical tasks, because the ECG is the most commonly used non-invasive examination for assessing cardiac condition. State-of-the-art algorithms rely on digital signal processing (DSP) techniques, which require considerable effort to re-tune rules for new morphologies. In contrast, deep learning (DL) algorithms, especially for classification, are attracting growing attention in academia and industry. This study proposes a three-stage hierarchical deep learning framework. In Stage 1, a one-dimensional U-Net performs coarse segmentation of the ECG signal to separate background from regions of cardiac activity. In Stage 2, an attention-fusion U-Net takes the probability map produced in Stage 1 to localize and classify the P wave, QRS complex and T wave more precisely in time. In Stage 3, complete heartbeat segments are extracted from the Stage 2 segmentation result and a beat-level classification model is trained to classify beats, following the AAMI EC57 standard, into normal beats (N) and three classes of abnormal rhythm (S, V and Q). The Stage 1 and Stage 2 segmentation models achieved Macro F1 scores of 0.965 and 0.999, respectively, indicating very high waveform localization accuracy. Combined with soft max post-processing, the final four-class beat classifier reached a Macro F1 of 0.9779 and an overall accuracy of 0.9970 on the test set, and remained stable on the minority classes (V and Q).
No takes yet. Share an insight, caveat, or question.
Chang Chia-Wei (2026) studied Cardiac arrhythmia and waveform segmentation in ECG recordings (n=105). Three-stage hierarchical U-Net and 1D-ResNet deep learning framework vs. Ground-truth cardiologist and reference beat annotations (.q1c and .atr files) was evaluated on Four-class heartbeat classification accuracy (AAMI EC57 standard: N, S, V, Q) on the test set. The hybrid U-Net and ResNet framework achieved 99.70% accuracy and a 0.9779 Macro F1-score for four-class ECG heartbeat recognition on the QT Database test set.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: