Deep Learning R-Wave Detection for Electrocardiographic Gating in Cardiac MRI

Area of Science:

• Medical Imaging
• Artificial Intelligence in Medicine
• Cardiovascular Imaging

Background:

• Electrocardiographic (ECG) gating is crucial for high-quality cardiac MRI.
• MRI-induced artifacts, particularly at 3.0 T, can lead to ECG-gating errors.
• Traditional signal processing algorithms may struggle with artifact-laden ECG signals.

Purpose of the Study:

• To quantify ECG-gating error frequency in cine cardiac MRI at 1.5 T and 3.0 T.
• To evaluate the efficacy of Convolutional Neural Networks (CNNs) in reducing these errors.

Main Methods:

• Retrospective analysis of ECG tracings from 120 cardiac MRI patients (1.5 T and 3.0 T) and an external dataset.
• Manual R-wave annotation for determining arrhythmia and ECG-gating error frequency.
• Development and testing of a CNN for R-wave detection, compared against VCG gating and the Hamilton algorithm.

Main Results:

• ECG-gating errors occurred in 8.1% of patients at 1.5 T and 15.2% at 3.0 T.
• CNNs achieved higher F1 scores than VCG (1.5 T) and the Hamilton algorithm (3.0 T).
• CNNs demonstrated a significantly lower false-positive rate at 3.0 T compared to the Hamilton algorithm.

Conclusions:

• CNNs provide a robust solution for R-wave detection in cardiac MRI, mitigating MRI-induced artifacts.
• The use of CNNs can significantly enhance ECG gating accuracy and improve overall cardiac MR image quality.
• Deep learning methods show promise for improving retrospective cardiac MRI reconstruction and quality.