Sequence to Sequence ECG Cardiac Rhythm Classification Using Convolutional Recurrent Neural Networks.

Area of Science:

• Cardiology
• Artificial Intelligence
• Signal Processing

Background:

• Accurate cardiac rhythm classification is crucial for diagnosing heart conditions.
• Existing methods for electrocardiogram (ECG) analysis face challenges with noise and data variability.

Purpose of the Study:

• To develop and validate a novel deep learning architecture for robust cardiac rhythm segmentation and classification.
• To process both ECG signal waveforms and spectrograms for improved classification accuracy.
• To assess the model's performance in the presence of label noise and its generalizability on external datasets.

Main Methods:

• A novel deep learning architecture combining Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) was developed.
• The model operates in a sequence-to-sequence framework, processing sliding windows of ECG signals.
• Input includes both ECG signal spectrograms and raw waveforms, enabling training with label noise.

Main Results:

• The proposed model achieved an average F1 score of 0.89 across 5 cardiac rhythm classes.
• Performance was validated on an independent external database, demonstrating generalizability.
• The architecture achieved comparable classification performance to state-of-the-art methods with fewer training parameters.

Conclusions:

• The novel deep learning architecture offers an effective and efficient approach for cardiac rhythm classification from ECG data.
• The model's ability to handle label noise and generalize to external data enhances its clinical applicability.
• This approach represents a significant advancement in automated ECG analysis for cardiology.