nucleAIzer: A Parameter-free Deep Learning Framework for Nucleus Segmentation Using Image Style Transfer

Hollandi, Réka; Szkalisity, Ábel; Tóth, Tímea; Tasnádi, Ervin; Molnár, Csaba; Mathe, Botond; Grexa, István; Molnár, J.; Bálind, Árpád; Gorbe, Mate; Kovács, Mária; Migh, Ede; Goodman, Allen; Balassa, Tamás; Koós, Krisztián; Wang, Wenyu; Caicedo, Juan C.; Bara, Norbert; Kovács, Ferenc; Paavolainen, Lassi; Danka, Tivadar; Kriston, András; Carpenter, Anne E.; Smith, Kevin; Horváth, Péter

doi:10.1016/j.cels.2020.04.003

Cited by 199 publications

(200 citation statements)

References 24 publications

Supporting

Mentioning

196

Contrasting

Unclassified

Order By: Relevance

“…These images are input to a nuclei segmentation pipeline, which flattens the images to white nuclei objects and black background. Nuclei images are segmented in 2D using the NucleAIzer platform maskRCNN Network, trained as described in Hollandi et al, 2020 . We trained the neural network with an expected nuclear radius of 32 pixels ( Figure 1—figure supplement 2D ).…”

Section: Methodsmentioning

confidence: 99%

The Wg and Dpp morphogens regulate gene expression by modulating the frequency of transcriptional bursts

Bakker

Mani

Carthew

2020

eLife

View full text Add to dashboard Cite

Morphogen signaling contributes to the patterned spatiotemporal expression of genes during development. One mode of regulation of signaling-responsive genes is at the level of transcription. Single-cell quantitative studies of transcription have revealed that transcription occurs intermittently, in bursts. Although the effects of many gene regulatory mechanisms on transcriptional bursting have been studied, it remains unclear how morphogen gradients affect this dynamic property of downstream genes. Here we have adapted single molecule fluorescence in situ hybridization (smFISH) for use in the Drosophila wing imaginal disc in order to measure nascent and mature mRNA of genes downstream of the Wg and Dpp morphogen gradients. We compared our experimental results with predictions from stochastic models of transcription, which indicated that the transcription levels of these genes appear to share a common method of control via burst frequency modulation. Our data help further elucidate the link between developmental gene regulatory mechanisms and transcriptional bursting.

show abstract

Section: Methodsmentioning

confidence: 99%

The Wg and Dpp morphogens regulate gene expression by modulating the frequency of transcriptional bursts

Bakker

Mani

Carthew

2020

eLife

View full text Add to dashboard Cite

show abstract

“…The quality of images created by GANs improves as the generator "learns" to fool the discriminator through an iterative process of trial and error. GANs have been applied in histopathology with success, including for stain translation from autofluorescence imaging [30], to removal of technical artifacts, as well as for nucleus detection and augmentation of deep learning datasets with synthetic images to improve prediction accuracy [13,[30][31][32][33][34][35][36][37][38][39][40][41][42].…”

Section: Introductionmentioning

confidence: 99%

A large-scale internal validation study of unsupervised virtual trichrome staining technologies on nonalcoholic steatohepatitis liver biopsies

et al. 2021

View full text Add to dashboard Cite

Non-alcoholic steatohepatitis (NASH) is a fatty liver disease characterized by accumulation of fat in hepatocytes with concurrent inflammation and is associated with morbidity, cirrhosis and liver failure. After extraction of a liver core biopsy, tissue sections are stained with hematoxylin and eosin (H&E) to grade NASH activity, and stained with trichrome to stage fibrosis. Methods to computationally transform one stain into another on digital whole slide images (WSI) can lessen the need for additional physical staining besides H&E, reducing personnel, equipment, and time costs. Generative adversarial networks (GAN) have shown promise for virtual staining of tissue. We conducted a large-scale validation study of the viability of GANs for H&E to trichrome conversion on WSI (n = 574). Pathologists were largely unable to distinguish real images from virtual/synthetic images given a set of twelve Turing Tests. We report high correlation between staging of real and virtual stains ($${\rho} = 0.86$$ ρ = 0.86 ; 95% CI: 0.84–0.88). Stages assigned to both virtual and real stains correlated similarly with a number of clinical biomarkers and progression to End Stage Liver Disease (Hazard Ratio HR = 2.06, 95% CI: 1.36–3.12, p < 0.001 for real stains; HR = 2.02, 95% CI: 1.40–2.92, p < 0.001 for virtual stains). Our results demonstrate that virtual trichrome technologies may offer a software solution that can be employed in the clinical setting as a diagnostic decision aid.

show abstract

“…Deep learning has revolutionised image processing (LeCun, Bengio, and Hinton 2015) . On specific tasks such as cell segmentation (Caicedo, Roth, et al 2019;Hollandi et al 2020) , cell classification (Cireşan et al 2013;Buggenthin et al 2017;Christian Matek et al 2019) or in-silico staining (Ounkomol et al 2018;Christiansen et al 2018) , deep learning algorithms have led to breakthroughs in biomedical image analysis. They now achieve higher accuracy than trained experts (Esteva et al 2017;McKinney et al 2020;Christian Matek et al 2019) and outperform humans at data processing speed and prediction consistency (Tschandl et al 2019;Liu et al 2019) .…”

Section: Introductionmentioning

confidence: 99%

InstantDL - An easy-to-use deep learning pipeline for image segmentation and classification

Waibel

Boushehri

Marr

2020

Preprint

View full text Add to dashboard Cite

AbstractMotivationDeep learning contributes to uncovering and understanding molecular and cellular processes with highly performant image computing algorithms. Convolutional neural networks have become the state-of-the-art tool to provide accurate, consistent and fast data processing. However, published algorithms mostly solve only one specific problem and they often require expert skills and a considerable computer science and machine learning background for application.ResultsWe have thus developed a deep learning pipeline called InstantDL for four common image processing tasks: semantic segmentation, instance segmentation, pixel-wise regression and classification. InstantDL enables experts and non-experts to apply state-of-the-art deep learning algorithms to biomedical image data with minimal effort. To make the pipeline robust, we have automated and standardized workflows and extensively tested it in different scenarios. Moreover, it allows to assess the uncertainty of predictions. We have benchmarked InstantDL on seven publicly available datasets achieving competitive performance without any parameter tuning. For customization of the pipeline to specific tasks, all code is easily accessible.Availability and ImplementationInstantDL is available under the terms of MIT licence. It can be found on GitHub: https://github.com/marrlab/InstantDLContactcarsten.marr@helmholtz-muenchen.de

show abstract

nucleAIzer: A Parameter-free Deep Learning Framework for Nucleus Segmentation Using Image Style Transfer

Abstract: Highlights d Robust method automatically adapting to various unseen experimental scenarios d Deep learning solution for accurate nucleus segmentation without user interaction d Accelerates, improves quality, and reduces complexity of bioimage analysis tasks

Cited by 199 publications

References 24 publications

The Wg and Dpp morphogens regulate gene expression by modulating the frequency of transcriptional bursts

The Wg and Dpp morphogens regulate gene expression by modulating the frequency of transcriptional bursts

A large-scale internal validation study of unsupervised virtual trichrome staining technologies on nonalcoholic steatohepatitis liver biopsies

InstantDL - An easy-to-use deep learning pipeline for image segmentation and classification

Contact Info

Product

Resources

About