Computational framework for generating large panoramic super-resolution images from localization microscopy

Abstract

Combining super-resolution localization microscopy with pathology creates new opportunities for biomedical researches. This combination requires a suitable image mosaic method for generating a panoramic image from many overlapping super-resolution images. However, current image mosaic methods are not suitable for this purpose. Here we proposed a computational framework and developed an image mosaic method called NanoStitcher. We generated ground truth datasets and defined criteria to evaluate this computational framework. We used both simulated and experimental datasets to prove that NanoStitcher exhibits better performance than two representative image mosaic methods. This study is helpful for the mature of super-resolution digital pathology.

Multiplexed and Millimeter-Scale Fluorescence Nanoscopy of Cells and Tissue Sections via Prism-Illumination and Microfluidics-Enhanced DNA-PAINT

“…We note that, at present, the stitched views are based on coarse alignments between matching gold fiduciaries in the overlapping regions. Optimal stitching with sub-diffractive registration precisions (on par with the image resolutions) requires that the optical aberrations are properly corrected first , and is currently under development. Imaging of the same 1 mm 2 FOVs would have taken 1–3 days if using DNA-PAINT with a standard 50 μm × 50 μm FOV or 6–8 h even with ASTER .…”

“…The latter two schemes also create homogeneous illumination across the FOV for more uniform imaging than the Gaussian beam used in this work. Image quality can also be improved by correcting for optical aberrations, which, even with optimized optics and light paths, cannot be neglected given the large FOVs. ,, Such aberrations also pose challenges when stitching images from multiple FOVs . Aside from these potential improvements, PRIME-PAINT is by design amenable to automation through integration of programmable fluidics, FOV stitching and buffer exchange, and online data processing. , Leveraging recent progress in multiplexing strategies − and 3D calibration over extended depth and large FOVs, , 3D PRIME-PAINT imaging of many more targets over several mm 2 sample areas should be readily feasible.…”

Large-field and fast super-resolution microscopic imaging method based on laser interferometry

“…Therefore, this technology has been extensively used in biomedical research and other fields. Now, biomedical researchers hope to obtain more sample information in a shorter time [14][15][16][17][18][19]. It is necessary to observe the sample microstructure clearly and collect its macroscopic information quickly.…”

Field dependent deep learning enables high-throughput whole-cell 3D super-resolution imaging

“…The trend has been pointed to developing high-throughput super-resolution imaging techniques for high content screening [3][4][5] . The typical strategy is for automated microscope to acquire small field of view (FOV) images one by one 6 and generate a mosaic image by post processing 7 . However, hardware automation is often not available in the typical microscopes, and some biological samples are not suitable to be volumetrically scanned.…”

“…However, hardware automation is often not available in the typical microscopes, and some biological samples are not suitable to be volumetrically scanned. Moreover, it is not easy to stitch the super resolution images with accuracy comparable to its high spatial resolution 7 . As most biological samples contain rich structural information in three dimensions, it becomes particularly challenging to obtain the whole-cell-scale 3D single-molecule-resolution image at high throughput.…”