Megapixel fluorescence microscopy through scattering layers made simple

A team from The Hebrew University of Jerusalem has introduced a new method for megapixel-scale fluorescence microscopy through complex scattering media. This approach resolves high-resolution images from several tens of widefield fluorescence-microscope frames without requiring specialized equipment such as spatial-light modulators or intensive computational processing. By efficiently correcting distortions caused by light scattering, the technique allows for clear imaging of dense and challenging targets. Its compatibility with conventional microscopy setups, coupled with the use of established matrix-based techniques, makes it practical for widespread use.

A recent study published in Science Advances, led by PhD student Gil Weinberg, MSc student Elad Sunray, and Prof. Ori Katz from the Institute of Applied Physics at The Hebrew University of Jerusalem, introduces a new approach to fluorescence imaging. This innovative method overcomes the detrimental effects of extreme light scattering in conventional fluorescence microscopy—one of the most crucial imaging techniques in life sciences. Supported by the European Research Council funding, the study unveils a high-resolution imaging technique capable of resolving complex scattering media, with promising applications in biological research, materials science, and beyond.

Random light scattering within or through dense and complex samples often hinders fluorescence imaging, leading to significant image distortion. While noninvasive coherent imaging through complex media has progressed in recent years, fluorescence imaging has remained limited by requirements for sparse targets, complex wavefront controls, or large data sets.

The researchers demonstrate megapixel-scale image reconstruction with fewer than 150 widefield fluorescence-microscope frames acquired under unknown random changing illuminations, all without using spatial light modulators (SLMs) or intensive computational resources. Its memory-efficient implementation drastically reduces computational demands, enabling the imaging of large and intricate samples. Unlike previous approaches, this technique does not depend on assumptions about object sparsity or requires managing low-order wavefront distortions.

Central to the approach is the construction of a 'virtual fluorescence-based reflection matrix,' an analog to the well-studied coherent reflection matrix in optics and ultrasound imaging, using a limited number of randomly illuminated frames. Once this mathematical equivalence is formulated, any of the well-established, powerful computational scattering-compensation techniques developed for coherent imaging can be applied to incoherent fluorescence imaging.

This advancement improves biological research, enabling clearer visualization of structures within dense tissues. Its compatibility with conventional microscopy setups enhances accessibility for both academic and industrial researchers, contributing to progress in optical imaging and providing new possibilities for exploring complex systems.