Dictionary-Based Reconstruction of Spatio-Temporal 3D Magnetic Field Images from Quantum Diamond Microscope

Abstract: Three-dimensional magnetic imaging with high spatio-temporal resolution is critical for probing current paths in various systems, from biosensing to microelectronics. Conventional 2D Fourier-based current source localization methods are ill-posed in multilayer or dynamic systems due to signal overlap and noise. In this work, we demonstrate an innovative nitrogen-vacancy (NV) center-based wide-field magnetic microscopy technique for dynamic three-dimensional imaging and localization of current sources. Using custom-fabricated multilayer micro-coil platform to emulate localized, time-varying currents similar to neuronal activity, we acquire magnetic field maps with micrometre-scale spatial and millisecond-scale temporal resolution using per-pixel lock-in-based detection. Source localization and image reconstruction are achieved using a Least Absolute Shrinkage and Selection Operator (LASSO)-based reconstruction framework that incorporates experimentally measured basis maps as spatial priors. Our method enables robust identification of active current sources across space and time, and significantly advances the accuracy of dynamic 3D current imaging and NV-based magnetometry for complex systems.