Complete optical phase recovery and quantitative phase imaging are performed instantly without a computer

The optical image and characterization of weak phase scattering objects, such as isolated cells, bacteria, and thin sections of tissue, often used in biological research and medical applications, have been of considerable interest for decades. Because of their optical properties, when these “phase objects” are illuminated by a light source, the amount of scattered light is usually much less than the light that passes directly through the sample, resulting in poor image contrast using traditional imaging techniques. This low contrast of the image can be overcome by using, for example, chemical stains or fluorescent labels. However, these methods of external labeling or staining are often tedious, expensive, and include toxic chemicals.

Quantitative phase Imaging (QPI) has emerged as a powerful unlabeled approach for optical research and sensing of various weakly scattering transparent phase objects. The last few decades have witnessed the development of numerous digital methods of quantitative phase imaging based on image reconstruction algorithms that run on computers to recover the phase image of an object from various interferometric measurements. These digital QPI techniques, which run from graphics processors (GPUs), have been used in a variety of applications, including pathology, cell biology, immunology, and cancer researchamong others.

In a new scientific paper published in Art Advanced optical materialsA team of optical engineers led by Professor Idagan Ozkan of the Department of Electrical and Computer Engineering and the California Institute of Nanosystems (CNSI) of the University of California, Los Angeles (UCLA), has developed a diffractive optical network for replacing the alcobe. in QPI systems with a series of passive optical surfaces that are spatially designed using deep learning. Unlike conventional QPI systems, where the phase recovery step is performed on a digital computer by measuring intensity or hologram, the QPI diffraction grid directly processes the optical waves generated by the object itself to obtain sample phase information in the form of light. propagated through a diffraction network. Thus, all phase recovery and quantitative phase imaging processes are completed at the speed of light and without the need for an external power source except light illumination. Once light interacts with the object of interest and propagates through spatially constructed passive layers, the reconstructed phase image of the sample appears at the output of the diffraction grid as an intensity image, successfully converting the phase features of the input object into an intensity image at the exit.

These results represent the first fully optical phase search and phase-to-intensity conversion achieved by diffraction. According to the results presented by the UCLA team, QPI diffraction networks trained using deep learning can not only generalize to invisible objects of the new phase, which are statistically similar to learning images, but also generalize to completely new types of objects with different spatial features . In addition, these QPI diffraction networks are designed so that the quantification of the input phase is invariant to possible changes in light intensity or image sensor detection efficiency. The team also showed that QPI diffraction networks can be optimized to preserve their quantitative phase image quality even with mechanical diffraction layer shifts.

The QPI diffraction networks reported by the UCLA team are a new concept of phase imaging that, in addition to its highest computational speed, completes the phase recovery process as light passes through thin, passive diffraction surfaces, and thus eliminates the energy consumption and memory usage required in digital QPI systems, potentially paving the way for a variety of new applications in microscopy and sensing.