Fig. 1. Workflow of a position-related deconvolution network for reconstructing large FOV images

Fig. 2. Large FOV images obtained by axial swept light-sheet microscopy. (a)(b) Images of fluorescent microspheres; (c)(d) images of mouse brain slices

Fig. 3. PSF used in this experiment and normalized intensity profiles along the dashed line

Fig. 4. Network training error curves

Fig. 5. Examples of artifact problem in DWDN on simulated datasets

Fig. 6. Comparison of X-resolution between raw and reconstruction images with a large FOV on the left, center, and right sides of the FOV

Fig. 7. Raw images and deconvolution results of neurons collected with the large FOV light-sheet microscopy. (a) Large FOV raw images, RRLB reconstruction results, and RL reconstruction results; (b) the reconstruction results of different deconvolution methods correspond to the orange dashed box located on the left side of the FOV in Fig.7 (a), partial enlarged images and corresponding normalized intensity profiles, with the profile position corresponding to the yellow dashed line in Fig.7 (b); (c) RRLB reconstruction results, raw image and RL reconstruction results, corresponding to the yellow dashed box located to the right side of the FOV in Fig.7 (a); (d) the reconstruction results of different deconvolution methods correspond to the blue dashed box in Fig.7 (c)

Table 1. PSNR of simulated dataset reconstruction results

Table 2. NCC of simulated dataset reconstruction results