Title : Super-resolution image reconstruction framework based on the joint spatial-frequency reconstruction method
Abstract:
Super-resolution structured illumination microscopy (SR-SIM) is finding increasing application in biomedical research due to its capacities of high-speed imaging and minimzing photodamage, which is superior capability to visualize the subcellular dynamics in living cells. However, an issue that has plagued SR-SIM since its inception, is an inefficient imaging process resulting from a disjointed workflow and slow, complex, reconstruction algorithms. This nullifies real-time imaging as 4 to 8 s is required to reconstruct a single SR image of 1024 × 1024 pixels in size and impedes the widespread application of SR-SIM among biologists. We have presented the Joint Spatial-Frequency Reconstruction method (JSFR-SIM), which recovers the super-resolution image in the spatial domain with a simplified workflow, and optimizes the image qualities in the frequency domain. Here we used this method as a framework to improve the application of SIM. To reduce the reconstruction artifacts, we firstly integrate the high-fidelity reconstruction algorithm (HiFi-SIM) based on PSF engineering with our JSFR method, ensuring high quality reconstruction results and improving reconstruction speed. We achieve real-time reconstruction, display and storage of high-fidelity, super-resolved images of live cells. Secondly, the framework is introduced into 3D-SIM and shorten the single-frame reconstruction time into 11 ms@1024×1024 pixels, providing a real-time imaging scheme for 3D super-resolution results preview. Finally, we extend the framework into nonlinear SIM, and compress the time into 8.7 ms@1536×1536 pixels for reconstructing a second-order nonlinear SIM data with 63 original images. These results demonstrate that our methods break through the traditional frequency domain reconstruction framework in SR-SIM and speed up the super-resolution image reconstruction. The JSFR-SIM will greatly improve the work efficiency of biologists and facilitate SR-SIM as a routine tool in more biomedical laboratories.
