Bayesian in situ point spread function modeling enables high-fidelity three-dimensional nanoscopy in deep tissue

Bayesian in situ PSF modeling (BIM) retrieves accurate specimen-matched PSF models from raw blinking data, enabling high-fidelity 3D single-molecule localization microscopy in deep tissue beyond one hundred micrometers.

• Why it matters: Deep tissue aberrations, scattering, and elevated background reduce the information content of single-molecule observations, creating a barrier to reliable in situ PSF estimation with deterministic mappings.
• What they did: BIM uses Bayesian uncertainty modeling of emitter axial position and posterior fusion to reconstruct PSFs directly from raw blinking data under native imaging conditions.
• The result: BIM resolves hollow mitochondrial membranes, amyloid-{beta} fibrils, dendrites, and spines with high fidelity, and with adaptive optics improves 3D reconstruction at depths exceeding one hundred micrometers.