# Large zoom ratio and adaptive aberration correction microscope using 4DPSF-aware Physical Degradation-guided Network

## TL;DR

A new microscope design enables large zoom ranges and corrects optical distortions in real-time using a combination of hardware and AI.

## Contribution

Proposes a joint optimization framework integrating optical design and a 4DPSF-aware neural network for adaptive aberration correction during zooming.

## Key findings

- The microscope achieves a zoom range of 10.6× to 101.4× with high image quality.
- The 4DPSF-PDNet algorithm effectively suppresses distortions and corrects dynamic aberrations.
- The system is promising for applications in biology, medicine, and materials science.

## Abstract

In response to the emerging demand for dynamic and cross-scale microscopic observation in fields such as biology, medicine, and materials science, the liquid lens has been widely adopted in modern microscopy to enable real-time, continuous optical zooming. However, the limited optical power of the liquid lens restricts the zoom range, and the nonlinear dynamic aberrations introduced during zooming can significantly degrade image quality. To address these challenges, a continuous optical zoom microscope with a large zoom ratio and adaptive aberration correction is proposed, based on an end-to-end joint optimization framework that integrates optical design and neural network guided by physical degradation. By incorporating spatially-variant, multi-wavelength, and continuous-magnification 4D PSF of the system as physical priors, this work achieves fast and high-quality continuous zoom imaging from 10.6× ~ 101.4×, while adaptively correcting complex aberrations that vary with both magnification and spatial location. The core hardware component of the system is a zoom objective lens with a movable relay image plane, which especially integrates electrowetting liquid lenses. On the algorithmic side, 4DPSF-aware Physical Degradation-guided Network (4DPSF-PDNet) is introduced for adaptive aberration correction during the zooming process. By embedding the PSF into the network, the method effectively suppresses distortions and artifacts and achieves precise correction of complex, dynamically varying aberrations. Experimental results demonstrate that the proposed adaptive continuous microscope holds significant promise for a wide range of applications in biology, medical diagnostics, and materials science.

A continuous optical zoom microscope based on an end-to-end joint optimization framework with large zoom ratio and adaptive aberration correction is proposed.