Identification of Abdominal ECG Lead Location Over Fetal Development for Non-invasive Single Channel W-NETR Fetal ECG Extractor

Recent research has focused on noninvasive techniques for capturing fetal ECGs (fECGs) using abdominal electrodes. These existing systems typically rely on a sophisticated and high number of abdominal ECG leads. The recently proposed W-NETR method addresses easy and real-time usability utilizing only a single non-invasive abdominal ECG lead using a deep learning approach. Despite its accuracy, its performance is influenced by precise lead placement, affecting its practical implementation. This study explores the optimal abdominal ECG lead placement for precise fECG extraction in early (20-22 weeks) and late (37-40 weeks) gestational periods. Using an 11-lead prototype by Atom Medical Co, we identify the best lead on the upper abdomen in the early stage and on the lower abdomen in the late stage, with a right back-placed reference electrode. We found that the single-leaded W-NETR pipeline not only accurately reconstructs fECG but also surpasses ground truth through enhanced noise filtration when optimal lead placement is used. This discovery opens avenues for implementing such lightweight models in future real-life fECG extractors, leveraging its computational efficiency and high performance.