Wavefront estimation through structured detection in laser scanning microscopy
Francesco Fersini,Alessandro Zunino,Pietro Morerio,Francesca Baldini,Martin J. Booth,Alessio Del Bue,Giuseppe Vicidomini
DOI: https://doi.org/10.1101/2024.12.31.630574
2025-01-01
Abstract:Laser scanning microscopy (LSM) serves as the backbone of numerous advanced imaging techniques, with confocal laser scanning microscopy (CLSM) being a pivotal tool in life sciences research. However, its effective resolution is often compromised by optical aberrations, a common challenge in all optical systems. While adaptive optics (AO) can correct these aberrations, current methods face significant limitations. Aberration estimation, which is central to any AO approach, typically requires specialized hardware or prolonged sample exposure, rendering these methods cumbersome, sample-invasive, and less user-friendly.
In this study, we present an efficient AO strategy for CLSM that used a detector array -- the same of super-resolved image-scanning microscopy -- with a standard AO element for light wavefront modulation. We demonstrate that imaging datasets acquired with a detector array inherently encode information about optical aberrations. Leveraging this property, we developed a custom convolutional neural network capable of decoding aberrations, up to the 11th Zernike coefficient, directly from a single acquisition. This method was validated using a custom image-scanning microscope equipped with a single-photon avalanche diode (SPAD) array detector.
This approach paves the way for a new generation of AO implementations in LSM, offering an accessible solution that minimizes sample stress while achieving high-resolution, and aberration-free imaging.
Bioengineering