Review of bio-optical imaging systems with a high space-bandwidth product

Abstract

Optical imaging has served as a primary method to collect information about biosystems across scales-from functionalities of tissues to morphological structures of cells and even at biomolecular levels. However, to adequately characterize a complex biosystem, an imaging system with a number of resolvable points, referred to as a space-bandwidth product (SBP), in excess of one billion is typically needed. Since a gigapixel-scale far exceeds the capacity of current optical imagers, compromises must be made to obtain either a low spatial resolution or a narrow field-of-view (FOV). The problem originates from constituent refractive optics-the larger the aperture, the more challenging the correction of lens aberrations. Therefore, it is impractical for a conventional optical imaging system to achieve an SBP over hundreds of millions. To address this unmet need, a variety of high-SBP imagers have emerged over the past decade, enabling an unprecedented resolution and FOV beyond the limit of conventional optics. We provide a comprehensive survey of high-SBP imaging techniques, exploring their underlying principles and applications in bioimaging.

Keywords: bioimaging; gigapixel imaging; high resolution; space-bandwidth product; wide field of view.

Fig. 1

The diffraction-limited SBP of standard microscope objective lenses at a 550 nm wavelength under incoherent illumination. The pathology slide image is modified from a public repository of image datasets (Image Data Resource).^,

Fig. 2

Array microscopy. (a) Images are captured through parallelized microimaging systems. (b) Schematics of an array microscopy for digital histopathology. In their system, three lenslet arrays are stacked. Each lens group has a diameter of 1.5 mm and a working distance of 400 μm. The microlenses are densely packed, and the orientation of the array is slightly tilted to the scanning axis. Therefore, a single-axis scan can provide the whole FOV. (c) Images of a pathology slide with a high SBP (∼10⁹). (d) Image of a fluorescently stained rat femur (upper) and its enlarged view (bottom) with parallelized scanning fluorescent microscopy. The scale bar for the top image is 1 mm, and the zoom-in images are 80 μm. (e) Sequential illumination of the beam for mechanical scanning-free parallel imaging. Panels (b), (c), (d), and (e) are modified from Refs. , , and , respectively.

Fig. 3

Multiscale optical systems. (a) Illustration of multiscale optical designs. (b) Schematic of the AWARE-2 camera consisting of multiscale optics and 98 microcameras. (c) The camera captures a 0.96 gigapixel image. (d) Multiscale optical system for bioimaging. The system can track traces of GFP-labeled immune cells. The scale bars are 1000 and 200 μm. Panels (b)–(d) are modified from Refs. , , and , respectively.

Fig. 4

High-SBP imaging with Fourier ptychography. (a) Principles of spatial frequency-domain multiplexing. (b) Simplified diagram of a phase-retrieval algorithm. (c) Recovery of the spatially varying pupil function. (d) High-resolution Fourier ptychography image of red blood cells. Particles are shown in the zoom-in view of malaria-infected red blood cells (arrow). Panels (c) and (d) are modified from Refs. and , respectively.

Fig. 5

Structured illumination microscopy. (a) Fourier domain representation of conventional, linear, and nonlinear structured illumination microscopy. In conventional microscopy, the measurable spatial frequency range is given as $|\overrightarrow{k}|>(2/\lambda )\cdot \text{NA}$. In linear structured illumination, spatial frequency information of the sample is laterally shifted an amount corresponding to the period of the illuminating pattern. Therefore, the high-spatial frequencies beyond the conventional imaging system become observable. In nonlinear structured illumination, the spatial frequency information of the sample is shifted corresponding to integer multiples of the pattern’s frequency. With pattern rotation, a large spatial frequency range can be collected. (b) A mammalian CHO cell imaged by the nonlinear structured illumination microscopy. Panel (b) is modified from Ref. .

Fig. 6

Hardware wavefront-engineering-based methods for high-SBP imaging. (a) System schematic of adaptive optical scanning microscopy. (b) The viewing location is given by the tilting angle of the galvanometric mirror, and the corresponding aberrations are corrected by the deformable mirror. (c) A bright-field image of a living C. elegans in a sub-FOV of the system. (d) Principles of high-resolution wide-FOV focusing with a disordered metasurface and wavefront shaping. (e) Scanning fluorescence microscopy with the metasurface. Immunofluorescence-labeled parasites (Giardia lamblia cysts) were imaged. The FOV and resolution are comparable to that of the 4×/0.1 NA objective lens and 20×/0.5 NA objective lens, respectively. Panels (a), (b)–(c), (d)–(e) are modified from Refs. , , and , respectively.

Fig. 7

Computational wavefront-engineering-based methods for high-SBP imaging. (a) Computational correction of aberrations in optical coherence tomography and interferometric synthetic aperture microscopy. (b) Computational correction of spatially varying aberrations of a wide-FOV objective lens (2×/0.08 NA). The system shows diffraction-limited performance over the entire FOV (13 mm in diameter). (c) Correcting spatially varying aberrations. The hardware approach sequentially corrects for aberrations at local positions. Correction of the whole FOV with averaged aberrations results in a degraded performance. By contrast, the computational approach can correct for spatially varying aberrations across the whole FOV without lateral scanning. Panels (a) and (b) are modified from Refs. and , respectively.

Fig. 8

Illustration of various high-SBP imaging techniques. The pathology slide image is modified from a public repository of image datasets (Image Data Resource).^,

Fig. 9

SBP of high-SBP imaging systems. We note that the cut-off spatial frequency of an incoherent imaging system is double that of the coherent imaging system given the same NA. All frequency-domain methods^,,, are coherent imaging methods. In this graph, the SBP values of objective lenses were calculated for incoherent imaging. For coherent imaging, the cut-off spatial frequency of the objective lenses will be halved. The number in the “Ref” column next to the author and year indicates the corresponding reference index.