Real-Time 3D Reconstruction of Nanomotor Dynamics Using Phase-Space Deconvolution Light-Field Microscopy

Quantitatively resolving nanoscale motion is essential for understanding colloidal active matter. Conventional 2D tracking, however, suffers from projection bias, which can lead to inaccurate diffusivity measurements and hinders accurate characterization of active propulsion. Here, we introduce a phase-space deconvolution light-field microscopy (LFM) system that enables scan-free, real-time volumetric imaging, allowing reconstruction of time-lapse 3D volumes and extraction of continuous 3D trajectories with millisecond temporal resolution and axial localization precision of ∼100 nm. We validate the platform using 200 nm colloidal particles, demonstrating that the measured 3D mean-squared displacement agrees with the Stokes-Einstein prediction within 0.61%-a sixfold improvement in diffusivity accuracy compared to conventional 2D tracking. When applied to enzyme-powered Janus nanomotors, our method quantitatively separates active propulsion from Brownian motion and reveals a ∼two-fold increase in effective diffusivity under glucose (D_eff = 5.97 µm² s^{- 1}). In living cells, it enables robust 3D tracking amid cellular autofluorescence and discriminates enzyme-powered nanomotors from passive controls. This framework establishes a robust approach for quantitative 3D characterization of both passive and active nanosystems, providing a direct experimental link between nanoscale dynamics and theoretical models of complex biological environments.