Bridging Cellular and Clinical Electrophysiology: Contactless Ventricular Action Potentials Reconstruction from Magnetocardiography, Quantum Sensors and AI

Magnetocardiography (MCG) can reconstruct the ventricular action potential waveform (VAPw) without contact, potentially bridging cellular and clinical electrophysiology. We assessed the reproducibility and physiological consistency of VAPw reconstruction from 36-channel unshielded SQUID MCG, using new proprietary custom software for automatic reconstruction and artifact editing. Fifteen healthy subjects were recorded twice at rest; ten healthy subjects and ten cardiac patients were recorded at rest and after bicycle exercise. The reconstructed duration at 80% repolarization (rVAPw80) was about three times more reproducible than the manually measured QT (standard error of measurement 3.4 vs 10.1 ms; intraclass correlation 0.97 vs 0.78), and its small test–retest differences reflected the signal rather than the placement of fiducial points. rVAPw80 shortened consistently with heart rate (within-subject slope 0.145 ms/ms in healthy subjects and 0.299 in patients; within-subject r 0.89 and 0.91), the shortening occurring before the T-wave peak, whereas reconstructed phase 0 was rate-independent in healthy subjects. In two noisy recordings, strong 50-Hz interference barely affected the reconstruction. Finally, we discuss how room-temperature quantum sensors, particularly nitrogen-vacancy (NV) diamond magnetometers, and artificial intelligence for fiducial-point detection, single-beat denoising and model-based reconstruction could bring non-invasive electrophysiology closer to the cellular scale.