Journal: Nature methods
Article Title: Deconwolf enables high-performance deconvolution of widefield fluorescence microscopy images.
doi: 10.1038/s41592-024-02294-7
Figure Lengend Snippet: Fig. 1 | Implementation and benchmarking of Deconwolf. a, Schematic Deconwolf workflow. WSI, whole-slide image. b,c, In silico generated microtubule images before (ground truth) (b) and after adding artificial noise to simulate a real image (c). Maximum z-projection is shown. d, MSE after deconvolving the image in c using the default Deconwolf mode with scaled heavy ball15 acceleration (DW_SHB), or Deconwolf based on the classic Richardson–Lucy deconvolution method (DW_RL)2,3. The dashed vertical lines indicate the number of iterations needed to reach the minimum MSE. e, As in c after deconvolution with Deconwolf (DW) using default settings. it, number of iterations. t, deconvolution time measured on an 8-Core AMD Ryzen 7 3700X machine. f, As in c using DeconvolutionLab2 (DL2) with default settings at 115 iterations. g, As in
Article Snippet: Visual inspection of the images in the original dataset showed densely packed clouds of fluorescent dots in different colors inside each nucleus, which could be only partially resolved by applying the commercial deconvolution software (Nikon NIS Elements AR, v5.02.0) incorporated in the OligoFISSEQ image processing pipeline (Fig. 6c).
Techniques: In Silico, Generated