Optical coherence tomography image denoising using a generative adversarial network with speckle modulation

Zhang, Dong; Liu, Guoyan; Ni, Guangming; Jerwick, Jason; Duan, Lian; Zhou, Chao

doi:10.1002/jbio.201960135

Cited by 58 publications

(31 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Obtained OCT datasets of the cardiac organoids were first processed to generate OCT images with corrected scales. Then OCT images were further de-noised using a speckle-modulation generative adversarial network 70 to reduce the speckle noise. 3D renderings of OCT images were performed using Amira software (Thermo Fisher Scientific).…”

Section: Methodsmentioning

confidence: 99%

Generation of Heart Organoids Modeling Early Human Cardiac Development Under Defined Conditions

Ball

Wasserman

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

Cardiovascular-related disorders are a significant worldwide health problem. Cardiovascular disease (CVD) is the leading cause of death in developed countries, making up a third of the mortality rate in the US 1 . Congenital heart defects (CHD) affect ~1% of all live births 2 , making it the most common birth defect in humans. Current technologies provide some insight into how these disorders originate but are limited in their ability to provide a complete overview of disease pathogenesis and progression due to their lack of physiological complexity. There is a pressing need to develop more faithful organ-like platforms recapitulating complex in vivo phenotypes to study human development and disease in vitro. Here, we report the most faithful in vitro organoid model of human cardiovascular development to date using human pluripotent stem cells (hPSCs). Our protocol is highly efficient, scalable, shows high reproducibility and is compatible with high-throughput approaches. Furthermore, our hPSC-based heart organoids (hHOs) showed very high similarity to human fetal hearts, both morphologically and in cell-type complexity. hHOs were differentiated using a two-step manipulation of Wnt signaling using chemical inhibitors and growth factors in completely defined media and culture conditions. Organoids were successfully derived from multiple independent hPSCs lines with very similar efficiency. hHOs started beating at ~6 days, were mostly spherical and grew up to ~1 mm in diameter by day 15 of differentiation. hHOs developed sophisticated, interconnected internal chambers and confocal analysis for cardiac markers revealed the presence of all major cardiac lineages, including cardiomyocytes (TNNT2 + ), epicardial cells (WT1 + , TJP + ), cardiac fibroblasts (THY1 + , VIM + ), endothelial cells (PECAM1 + ), and endocardial cells (NFATC1 + ). Morphologically, hHOs developed well-defined epicardial and adjacent myocardial regions and presented a distinct vascular plexus as well as endocardial-lined microchambers. RNA-seq time-course analysis of hHOs, monolayer differentiated iPSCs and fetal human hearts revealed that hHOs recapitulate human fetal heart tissue development better than previously described differentiation protocols 3,4 . hHOs allow higher-order interaction of distinct heart tissues for the first time and display biologically relevant physical and topographical 3D cues that closely resemble the human fetal heart. Our model constitutes a powerful novel tool for discovery and translational studies in human cardiac development and disease.

show abstract

Section: Methodsmentioning

confidence: 99%

Generation of Heart Organoids Modeling Early Human Cardiac Development Under Defined Conditions

Ball

Wasserman

et al. 2020

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…EstimatingX from X is an inverse problem and the challenge is how to model the function that is able to generate E, so that X can be estimated. Many studies have been conducted to estimate E. Recently, GAN has commonly used and proven its effectively to solve this task [28][29][30][31][32][33].…”

Section: Proposed Methodsmentioning

confidence: 99%

Background Load Denoising across Complex Load Based on Generative Adversarial Network to Enhance Load Identification

Mukaroh

Kim

2020

Sensors

View full text Add to dashboard Cite

Non-Intrusive Load Monitoring (NILM) allows load identification of appliances through a single sensor. By using NILM, users can monitor their electricity consumption, which is beneficial for energy efficiency or energy saving. In advance NILM systems, identification of appliances on/off events should be processed instantly. Thus, it is necessary to use an extremely short period signal of appliances to shorten the time delay for users to acquire event information. However, acquiring event information from a short period signal raises another problem. The problem is target load feature to be easily mixed with background load. The more complex the background load has, the noisier the target load occurs. This issue certainly reduces the appliance identification performance. Therefore, we provide a novel methodology that leverages Generative Adversarial Network (GAN) to generate noise distribution of background load then use it to generate a clear target load. We also built a Convolutional Neural Network (CNN) model to identify load based on single load data. Then we use that CNN model to evaluate the target load generated by GAN. The result shows that GAN is powerful to denoise background load across the complex load. It yields a high accuracy of load identification which could reach 92.04%.

show abstract

“…( a ) Example automatic retinal layer segmentation using DL compared to manual segmentation (reprinted from [174]). ( b ) GAN for denoising OCT images (adapted from [180]). (c) Attention map overlaid with retinal images indicated features that CNN used for diagnosing normal versus age‐related macular degeneration (AMD) [181] (Reproduced from Rim et al [181], with permission from BMJ Publishing Group Ltd.).…”

Section: Applications In Biomedical Opticsmentioning

confidence: 99%

Deep Learning in Biomedical Optics

et al. 2021

View full text Add to dashboard Cite

This article reviews deep learning applications in biomedical optics with a particular emphasis on image formation. The review is organized by imaging domains within biomedical optics and includes microscopy, fluorescence lifetime imaging, in vivo microscopy, widefield endoscopy, optical coherence tomography, photoacoustic imaging, diffuse tomography, and functional optical brain imaging. For each of these domains, we summarize how deep learning has been applied and highlight methods by which deep learning can enable new capabilities for optics in medicine. Challenges and opportunities to improve translation and adoption of deep learning in biomedical optics are also summarized. Lasers Surg. Med. © 2021 Wiley Periodicals LLC

show abstract

Optical coherence tomography image denoising using a generative adversarial network with speckle modulation

Cited by 58 publications

References 42 publications

Generation of Heart Organoids Modeling Early Human Cardiac Development Under Defined Conditions

Generation of Heart Organoids Modeling Early Human Cardiac Development Under Defined Conditions

Background Load Denoising across Complex Load Based on Generative Adversarial Network to Enhance Load Identification

Deep Learning in Biomedical Optics

Contact Info

Product

Resources

About