Imaging flow cytometry

Nature Reviews Methods Primers, Journal Year: 2022, Volume and Issue: 2(1)

Published: Nov. 3, 2022

Language: Английский

---

Artificial Intelligence of Things (AIoT) Advances in Aquaculture: A Review

Processes, Journal Year: 2025, Volume and Issue: 13(1), P. 73 - 73

Published: Jan. 1, 2025

The integration of artificial intelligence (AI) and the internet things (IoT), known as (AIoT), is driving significant advancements in aquaculture industry, offering solutions to longstanding challenges related operational efficiency, sustainability, productivity. This review explores latest research studies AIoT within focusing on real-time environmental monitoring, data-driven decision-making, automation. IoT sensors deployed across systems continuously track critical parameters such temperature, pH, dissolved oxygen, salinity, fish behavior. AI algorithms process these data streams provide predictive insights into water quality management, disease detection, species identification, biomass estimation, optimized feeding strategies, among others. Much adoption advantageous various fronts, there are still numerous challenges, including high implementation costs, privacy concerns, need for scalable adaptable models diverse environments. also highlights future directions aquaculture, emphasizing potential hybrid models, improved scalability large-scale operations, sustainable resource management.

Language: Английский

---

Machine learning for microalgae detection and utilization

Frontiers in Marine Science, Journal Year: 2022, Volume and Issue: 9

Published: July 26, 2022

Microalgae are essential parts of marine ecology, and they play a key role in species balance. also have significant economic value. However, microalgae too tiny, there many different kinds single drop seawater. It is challenging to identify monitor changes. Machine learning techniques achieved massive success object recognition classification, attracted wide range attention. Many researchers introduced machine algorithms into applications, similarly effects gained. The paper summarizes recent advances based on various such as bioenergy generation from microalgae, environment purification with growth monitor. Finally, we prospect development treatment the future.

Language: Английский

---

Survey of automatic plankton image recognition: challenges, existing solutions and future perspectives

Artificial Intelligence Review, Journal Year: 2024, Volume and Issue: 57(5)

Published: April 12, 2024

Abstract Planktonic organisms including phyto-, zoo-, and mixoplankton are key components of aquatic ecosystems respond quickly to changes in the environment, therefore their monitoring is vital follow understand these changes. Advances imaging technology have enabled novel possibilities study plankton populations, but manual classification images time consuming expert-based, making such an approach unsuitable for large-scale application urging automatic solutions analysis, especially recognizing species from images. Despite extensive research done on recognition, latest cutting-edge methods not been widely adopted operational use. In this paper, a comprehensive survey existing recognition presented. First, we identify most notable challenges that make development systems difficult restrict deployment Then, provide detailed description found literature. Finally, propose workflow specific new datasets recommended approaches address them. Many important remain unsolved following: (1) domain shift between hindering instrument independent system, (2) difficulty process previously unseen classes non-plankton particles, (3) uncertainty expert annotations affects training machine learning models. To build harmonized location agnostic purposes should be addressed future research.

Language: Английский

---

Pollen analysis using multispectral imaging flow cytometry and deep learning

New Phytologist, Journal Year: 2020, Volume and Issue: 229(1), P. 593 - 606

Published: Aug. 28, 2020

Summary Pollen identification and quantification are crucial but challenging tasks in addressing a variety of evolutionary ecological questions (pollination, paleobotany), also for other fields research (e.g. allergology, honey analysis or forensics). Researchers exploring alternative methods to automate these but, several reasons, manual microscopy is still the gold standard. In this study, we present new method pollen using multispectral imaging flow cytometry combination with deep learning. We demonstrate that our allows fast measurement while delivering high accuracy identification. A dataset 426 876 images depicting from 35 plant species was used train convolutional neural network classifier. found best‐performing classifier yield species‐averaged 96%. Even difficult differentiate could be clearly separated. Our approach detailed determination morphological traits, such as size, symmetry structure. phylogenetic analyses suggest conservatism some traits. Given comprehensive reference database, provide powerful tool any study need rapid accurate identification, grain trait extraction recent pollen.

Language: Английский

---

Automatic plankton quantification using deep features

Journal of Plankton Research, Journal Year: 2019, Volume and Issue: 41(4), P. 449 - 463

Published: May 11, 2019

Abstract The study of marine plankton data is vital to monitor the health world’s oceans. In recent decades, automatic recognition systems have proved useful address vast amount collected by specially engineered in situ digital imaging systems. At beginning, these were developed and put into operation using traditional classification techniques, which fed with hand-designed local image descriptors (such as Fourier features), obtaining quite successful results. past few years, there been many advances computer vision community rebirth neural networks. this paper, we leverage how computed convolutional networks trained out-of-domain are replace task estimating prevalence each class a water sample. To achieve goal, designed broad set experiments that show effective deep features when working combination state-of-the-art quantification algorithms.

Language: Английский

---

Towards operational phytoplankton recognition with automated high-throughput imaging, near-real-time data processing, and convolutional neural networks

Frontiers in Marine Science, Journal Year: 2022, Volume and Issue: 9

Published: Sept. 2, 2022

Plankton communities form the basis of aquatic ecosystems and elucidating their role in increasingly important environmental issues is a persistent research question. Recent technological advances automated microscopic imaging, together with cloud platforms for high-performance computing, have created possibilities collecting processing detailed high-frequency data on planktonic communities, opening new horizons testing core hypotheses ecosystems. Analyzing continuous streams big calls development deployment novel computer vision machine learning systems. The implementation these analysis systems not always straightforward regards to operationality, regarding flows, computing treatment need be considered. We pipeline near-real-time classification phytoplankton during remote imaging flow cytometer (Imaging FlowCytobot, IFCB). Convolutional neural network (CNN) used classify probability thresholds filter out images belonging our existing classes. system were monitor dominating species filamentous cyanobacteria coast Finland summer 2021. demonstrate that good recognition can achieved transfer utilizing relatively shallow, publicly available, pre-trained CNN model fine-tuning it community-specific (overall F1-score 0.95 test set labeled image complemented 50% unclassifiable portion). This enables both fast training low resource requirements making easy modify applicable wide range situations. performed well when natural community over different seasons 0.82 evaluation set). Furthermore, we address key challenges varying analyze practical implications confused published Baltic Sea models (~63000 50 classes) accelerate other brackish freshwater communities. Our set, 59 fully annotated samples throughout an annual cycle, also available purposes (~150000 images).

Language: Английский

---

Smart systems in producing algae-based protein to improve functional food ingredients industries

Food Research International, Journal Year: 2023, Volume and Issue: 165, P. 112480 - 112480

Published: Jan. 13, 2023

Language: Английский

---

Deep Learning‐Based Single‐Cell Optical Image Studies

Cytometry Part A, Journal Year: 2020, Volume and Issue: 97(3), P. 226 - 240

Published: Jan. 25, 2020

Abstract Optical imaging technology that has the advantages of high sensitivity and cost‐effectiveness greatly promotes progress nondestructive single‐cell studies. Complex cellular image analysis tasks such as three‐dimensional reconstruction call for machine‐learning in cell optical research. With rapid developments high‐throughput flow cytometry, big data images are always obtained may require machine learning analysis. In recent years, deep been prevalent field large‐scale processing analysis, which brings a new dawn studies with an explosive growth availability. Popular techniques offer ideas multimodal multitask This article provides overview basic knowledge its applications We explore feasibility applying to where popular transfer learning, end‐to‐end have reviewed. Image preprocessing model training methods then summarized. Applications based on reviewed, include segmentation, super‐resolution reconstruction, tracking, counting, cross‐modal design control systems. addition, label‐free imaging, high‐content screening, cytometry also mentioned. Finally, perspectives discussed. © 2020 International Society Advancement Cytometry

Language: Английский

---

Deep learning-based diatom taxonomy on virtual slides

Scientific Reports, Journal Year: 2020, Volume and Issue: 10(1)

Published: Sept. 2, 2020

Abstract Deep convolutional neural networks are emerging as the state of art method for supervised classification images also in context taxonomic identification. Different morphologies and imaging technologies applied across organismal groups lead to highly specific image domains, which need customization deep learning solutions. Here we provide an example using (CNNs) identification morphologically diverse microalgal group diatoms. Using a combination high-resolution slide scanning microscopy, web-based collaborative annotation diatom-tailored analysis, assembled diatom database from two Southern Ocean expeditions. We use these data investigate effect CNN architecture, background masking, set size possible concept drift upon performance. Surprisingly, VGG16, relatively old network showed best performance generalizing ability on our images. previous study, found that masking slightly improved In general, training only classifier top layers pre-trained extensive, but not domain-specific surprisingly high (F1 scores around 97%) with already few (100–300) examples per class, indicating domain adaptation novel can be feasible limited investment effort.

Language: Английский

---