Deep learning-based fast multispectral Fourier ptychographic microscopy

doi:10.1016/j.optlastec.2025.113464

Deep learning-based fast multispectral Fourier ptychographic microscopy

2025-07-04

0

PRE

AI

DOI：10.1016/j.optlastec.2025.113464

Original

Original request for help

Save

Abstract

En 中文

• A deep learning-based fast multispectral Fourier ptychographic microscopy (FMSFPM) framework is proposed, enabling direct multispectral reconstruction from RGB outputs without additional spectral acquisition. • The method employs a modified U-Net with attention mechanisms to enhance spectral-spatial feature learning, achieving superior performance compared to state-of-the-art models on public and microscopic datasets. • FMSFPM significantly improves imaging throughput and spectral fidelity, offering a compact and scalable solution for biomedical and optical applications.

Deep learning-based fast multispectral Fourier ptychographic microscopy

Abstract

Journal

Organization