Aug 02, 2023
Label-free Quantitative Phase Imaging (QPI) offers non-toxic, high-resolution imaging of transparent specimens, avoiding complex sample preparation. It's a valuable tool in biomedical sciences for studying weakly scattering phase objects like cells.
Conventional QPI systems are slow and resource-intensive with complex digital reconstruction and phase retrieval algorithms. Many methods neglect random scattering in biological tissue, posing limitations.
In Light: Advanced Manufacturing, Prof. Aydogan Ozcan's UCLA team presents a novel quantitative phase imaging technique for objects obscured by random phase diffusers. Their method employs a deep learning-based diffractive optical network spanning ~70λ.
The training incorporated diverse random phase diffusers to enhance resilience against unknown perturbations. Once trained, the diffractive layers enable all-optical phase recovery and quantitative imaging of completely hidden objects by random diffusers.
Numerical simulations proved the QPI diffractive network's ability to image new objects with unseen random phase diffusers. They explored factors like spatially-structured layers and trade-offs between quality and efficiency, favoring deeper networks. The system can scale across electromagnetic spectrum regions without layer redesign or retraining.
The all-optical computing framework offers advantages like low power usage, high frame rate, and compact size. The UCLA team envisions integrating QPI diffractive designs onto image sensor chips, creating a diffractive QPI microscope with on-chip phase recovery and image reconstruction through light diffraction within passive layers.
Visit our website and learn more!