Incorporating the image formation process into deep learning improves network performance | Nature Methods
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript. Advertisement Nature Methods volume 19, pages 1427–1437 (2022)Cite this article 19k Accesses 26 Citations 72 Altmetric Metrics details We present Richardson–Lucy network (RLN), a fast and lightweight deep learning method for three-dimensional fluorescence microscopy deconvolution. RLN combines the traditional Richardson–Lucy iteration with a fully convolutional network structure, establishing a connection to the image formation process and thereby improving network performance. Containing only roughly 16,000 parameters, RLN enables four- to 50-fold faster processing than purely data-driven networks with many more parame
Main All fluorescence images are contaminated by blurring and noise, but this degradation can be ameliorated with deconvolution1,2,3. For example, iterative Richardson–Lucy deconvolution (RLD)4,5 is commonly used in fluorescence microscopy, and is appropriate if the dominant noise source is described by a Poisson distribution. Unfortunately, RLD is computationally taxing for three-dimensional (3D) and 3D timelapse (4D) data, particularly if complex regularization6,7 or large numbers of iterations are applied. To address this challenge, we recently proposed RLD variants8 that can accelerate…
saved by
related reading
- Aman's AI Journal • Primers • Ilya Sutskever's Top 30aman.ai
- ImageNet Classification with Deep Convolutional Neural Networksproceedings.neurips.cc
- Ilya 30u30arc.net
- The Little Book of Deep Learningfleuret.org
- The Decade of Deep Learning | Leo Gaobmk.sh
- [1505.04597] U-Net: Convolutional Networks for Biomedical Image Segmentationarxiv.org
- 1512.03385arxiv.org
- CS231n Deep Learning for Computer Visioncs231n.github.io
- Unsupervised Learning with SURE.pdfarxiv.org
- RAFT: Recurrent All-Pairs Field Transforms for Optical Flow | PDFarxiv.org
- Convolutional Neural Networks, Explained | Towards Data Sciencetowardsdatascience.com
- deconvolution_notes.pdfcs.toronto.edu