U-Net: deep learning for cell counting, detection, and morphometry

Nature Methods, Journal Year: 2018, Volume and Issue: 16(1), P. 67 - 70

Published: Dec. 4, 2018

Language: Английский

---

Deep learning for cellular image analysis

Nature Methods, Journal Year: 2019, Volume and Issue: 16(12), P. 1233 - 1246

Published: May 27, 2019

Language: Английский

---

Noise2Void - Learning Denoising From Single Noisy Images

2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Journal Year: 2019, Volume and Issue: unknown, P. 2124 - 2132

Published: June 1, 2019

The field of image denoising is currently dominated by discriminative deep learning methods that are trained on pairs noisy input and clean target images. Recently it has been shown such can also be without targets. Instead, independent images used, in an approach known as Noise2Noise (N2N). Here, we introduce Noise2Void (N2V), a training scheme takes this idea one step further. It does not require pairs, nor Consequently, N2V allows us to train directly the body data denoised therefore applied when other cannot. Especially interesting application biomedical data, where acquisition targets, or noisy, frequently possible. We compare performance approaches have either and/or available. Intuitively, cannot expected outperform more information available during training. Still, observe drops moderation compares favorably training-free methods.

Language: Английский

---

Inference in artificial intelligence with deep optics and photonics

Nature, Journal Year: 2020, Volume and Issue: 588(7836), P. 39 - 47

Published: Dec. 2, 2020

Language: Английский

---

Single-molecule localization microscopy

Nature Reviews Methods Primers, Journal Year: 2021, Volume and Issue: 1(1)

Published: June 3, 2021

Single-molecule localization microscopy (SMLM) describes a family of powerful imaging techniques that dramatically improve spatial resolution over standard, diffraction-limited and can image biological structures at the molecular scale. In SMLM, individual fluorescent molecules are computationally localized from sequences localizations used to generate super-resolution or time course images, define trajectories. this Primer, we introduce basic principles SMLM before describing main experimental considerations when performing including labelling, sample preparation, hardware requirements acquisition in fixed live cells. We then explain how low-resolution processed reconstruct images and/or extract quantitative information, highlight selection discoveries enabled by closely related methods. discuss some limitations potential artefacts as well ways alleviate them. Finally, present an outlook on advanced promising new developments fast-evolving field SMLM. hope Primer will be useful reference for both newcomers practitioners

Language: Английский

---

Label-free prediction of three-dimensional fluorescence images from transmitted-light microscopy

Nature Methods, Journal Year: 2018, Volume and Issue: 15(11), P. 917 - 920

Published: Sept. 6, 2018

Language: Английский

---

DeepPoseKit, a software toolkit for fast and robust animal pose estimation using deep learning

eLife, Journal Year: 2019, Volume and Issue: 8

Published: Oct. 1, 2019

Quantitative behavioral measurements are important for answering questions across scientific disciplines-from neuroscience to ecology. State-of-the-art deep-learning methods offer major advances in data quality and detail by allowing researchers automatically estimate locations of an animal's body parts directly from images or videos. However, currently available animal pose estimation have limitations speed robustness. Here, we introduce a new easy-to-use software toolkit, DeepPoseKit, that addresses these problems using efficient multi-scale model, called Stacked DenseNet, fast GPU-based peak-detection algorithm estimating keypoint with subpixel precision. These improve processing >2x no loss accuracy compared methods. We demonstrate the versatility our multiple challenging tasks laboratory field settings-including groups interacting individuals. Our work reduces barriers advanced tools measuring behavior has broad applicability sciences.

Language: Английский

---

Democratising deep learning for microscopy with ZeroCostDL4Mic

Nature Communications, Journal Year: 2021, Volume and Issue: 12(1)

Published: April 15, 2021

Abstract Deep Learning (DL) methods are powerful analytical tools for microscopy and can outperform conventional image processing pipelines. Despite the enthusiasm innovations fuelled by DL technology, need to access compatible resources train networks leads an accessibility barrier that novice users often find difficult overcome. Here, we present ZeroCostDL4Mic, entry-level platform simplifying leveraging free, cloud-based computational of Google Colab. ZeroCostDL4Mic allows researchers with no coding expertise apply key perform tasks including segmentation (using U-Net StarDist), object detection YOLOv2), denoising CARE Noise2Void), super-resolution Deep-STORM), image-to-image translation Label-free prediction - fnet, pix2pix CycleGAN). Importantly, provide suitable quantitative each network evaluate model performance, allowing optimisation. We demonstrate application study multiple biological processes.

Language: Английский

---

Applications, promises, and pitfalls of deep learning for fluorescence image reconstruction

Nature Methods, Journal Year: 2019, Volume and Issue: 16(12), P. 1215 - 1225

Published: July 8, 2019

Language: Английский

---

Glioblastoma hijacks neuronal mechanisms for brain invasion

Cell, Journal Year: 2022, Volume and Issue: 185(16), P. 2899 - 2917.e31

Published: July 31, 2022

Language: Английский

---