Time-Reversal Matrix Speeds Volumetric Adaptive Optical Imaging

Researchers in South Korea have described an imaging technique that enables high-throughput, 3D adaptive optical imaging of living samples, potentially leading to faster diagnosis and advances in neuroscience research.

The new method builds on reflection matrix microscopy, which was previously described by researchers at Institute of Basic Science (IBS) as a means to overcome the background noise seen in conventional microscopy caused by scattering as light travels through a tissue.

Reflection matrix microscopy combines hardware and computational advanced adaptive optics (AO) to measure a reflection matrix and enable undistorted images to be extracted from a matrix via post-digital image processing. The method emerged as a means to carry out label-free, non-invasive, high-resolution optical imaging deep inside biological tissues, but wasn’t without drawbacks. Measuring the entire reflection matrix is time-consuming and vulnerable to external perturbations because a large number of interferometric images for all accessible input illumination fields needs to be measured. While sparser sampling can speed up the process, insufficient sampling can lead to the limited capability to correct distortions. This means that real-time volumetric imaging of living samples was not possible, which led to practical limitations in its application to biodynamic studies.

In new research published in the journal Light: Science & Applications, the same IBS group unveiled a version of this AO technology that enables 3D imaging over a wide depth range in highly aberrated samples while minimizing image degradation. To speed up data acquisition, Prof. Choi Wonshik from the Center for Molecular Spectroscopy and Dynamics (CMSD) at IBS and his colleagues used a compressed sensing technique in the context of matrix imaging.

By adding a rotating optical diffuser in their previously deployed reflection matrix microscope, they were able to sequentially illuminate unknown speckle patterns on a sample. They then obtained a compressed reflection matrix by taking a much smaller number of speckle images than previously required, which reduced the matrix acquisition time by nearly 100 times. In image post-processing, they employed a compressed “time-reversal matrix” and a unique algorithm to identify sample information and aberrations separately. This technique dramatically reduces matrix acquisition time and eliminates the need for careful calibration or specific selection of illumination patterns.

The authors demonstrated their new microscope’s capabilities by achieving aberration-free 3D imaging of myelin nerve fibers in a mouse brain. The data acquisition time was only 3.58 seconds for volumetric imaging of 128 × 128 × 125 μm³ tissue, with a diffraction-limited lateral resolution of 0.45 μm and an axial resolution of 2 μm.

Choi and coauthors expect the method to open a new avenue for the practical application of matrix imaging in all fields of wave engineering, including biomedical imaging.  “Faster reflection matrix imaging technology is expected to enable real-time, nondestructive 3D optical diagnosis in the future, which will lead to faster diagnosis and advances in neuroscience research,” Dr. Choi says. “We will further develop it to broaden the scope of its application in all wave engineering disciplines, including biomedical imaging.”