Advanced ECG heart age estimation applicable to both sinus and non-sinus rhythm associates with cardiovascular risk, cardiovascular morbidity, and survival DOI Creative Commons
Zaidon Al-Falahi, Todd T. Schlegel,

Israel Lamela-Palencia

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2024, Volume and Issue: unknown

Published: March 13, 2024

Background An explainable advanced electrocardiography (A-ECG) heart age gap is the difference between A-ECG and chronological age. This an estimate of accelerated cardiovascular ageing expressed in years healthy human aging, can intuitively communicate risk to general population. However, existing measures require discernible P waves on ECG. Aims To develop prognostically validate a revised, without incorporating P-wave measures. Methods (non-P) was derived from 10-second 12-lead ECG derivation cohort using multivariable regression Bayesian 5-minute as reference. The non-P externally validated separate patients referred for magnetic resonance imaging by describing its association with failure hospitalization or death Cox regression, comorbidities. Results In (n=2771), agreed 5-min (R 2 =0.91, bias 0.0±6.7 years), increased increasing co-morbidity. validation (n=731, mean 54±15 years, 43% female, n=139 events over 5.7 [4.8–6.7] follow-up), (≥10 years) associated (hazard ratio [95% confidence interval] 2.04 [1.38–3.00], C-statistic 0.58 [0.54–0.62], presence hypertension, diabetes mellitus, hypercholesterolemia, (p≤0.009 all). Conclusions applicable both sinus non-sinus rhythm associates risk, morbidity, survival.

Language: Английский

Advancements in Deep Learning for B-Mode Ultrasound Segmentation: A Comprehensive Review DOI
Mohammed Yusuf Ansari, Iffa Afsa Changaai Mangalote, Pramod Kumar Meher

et al.

IEEE Transactions on Emerging Topics in Computational Intelligence, Journal Year: 2024, Volume and Issue: 8(3), P. 2126 - 2149

Published: April 2, 2024

Ultrasound (US) is generally preferred because it of low-cost, safe, and non-invasive. US image segmentation crucial in analysis. Recently, deep learning-based methods are increasingly being used to segment images. This survey systematically summarizes highlights aspects the learning techniques developed last five years for various body regions. We investigate analyze most popular loss functions metrics training evaluating neural network segmentation. Furthermore, we study patterns architectures proposed regions interest. present modules priors that address anatomical challenges associated with different organs have found variants U-Net dedicated overcome low-contrast blurry nature images suitable Finally, also discuss advantages context

Language: Английский

Citations

44
Utilizing machine learning for predicting drug release from polymeric drug delivery systems DOI Creative Commons

Sareh Aghajanpour,

Hamid Amiriara, Mehdi Esfandyari‐Manesh

et al.

Computers in Biology and Medicine, Journal Year: 2025, Volume and Issue: 188, P. 109756 - 109756

Published: Feb. 19, 2025

Language: Английский

Citations

2
Enhancing ECG-based heart age: impact of acquisition parameters and generalization strategies for varying signal morphologies and corruptions DOI Creative Commons
Mohammed Yusuf Ansari, Marwa Qaraqe, Raffaella Righetti

et al.

Frontiers in Cardiovascular Medicine, Journal Year: 2024, Volume and Issue: 11

Published: July 4, 2024

Electrocardiogram (ECG) is a non-invasive approach to capture the overall electrical activity produced by contraction and relaxation of cardiac muscles. It has been established in literature that difference between ECG-derived age chronological represents general measure cardiovascular health. Elevated strongly correlates with conditions (e.g., atherosclerotic disease). However, neural networks for ECG estimation are yet be thoroughly evaluated from perspective acquisition parameters. Additionally, deep learning systems analysis encounter challenges generalizing across diverse morphologies various ethnic groups susceptible errors signals exhibit random or systematic distortions To address these challenges, we perform comprehensive empirical study determine threshold sampling rate duration while considering their impact on computational cost networks. tackle concern waveform variability different populations, evaluate feasibility utilizing pre-trained fine-tuned estimate groups. empirically demonstrate finetuning an environmentally sustainable way train networks, it significantly decreases instances required (by more than 100× ) attaining performance similar trained weight initialization complete dataset. Finally, systematically augmentation schemes context introduce cropping scheme provides best-in-class using shorter-duration signals. The results also show enables well signal corruptions.

Language: Английский

Citations

15
Advancing paleontology: a survey on deep learning methodologies in fossil image analysis DOI Creative Commons
Mohammed Yaqoob Ansari, Mohammed Ishaq Mohammed, Mohammed Yusuf Ansari

et al.

Artificial Intelligence Review, Journal Year: 2025, Volume and Issue: 58(3)

Published: Jan. 6, 2025

Abstract Understanding ancient organisms and their interactions with paleoenvironments through the study of body fossils is a central tenet paleontology. Advances in digital image capture now allow for efficient accurate documentation, curation, interrogation fossil forms structures two three dimensions, extending from microfossils to larger specimens. Despite these developments, key processing analysis tasks, such as segmentation classification, still require significant user intervention, which can be labor-intensive subject human bias. Recent advances deep learning offer potential automate analysis, improving throughput limiting operator emergence within paleontology last decade, challenges scarcity diverse, high quality datasets complexity morphology necessitate further advancement will aided by adoption concepts other scientific domains. Here, we comprehensively review state-of-the-art based methodologies applied grouping studies on type nature task. Furthermore, analyze existing literature tabulate dataset information, neural network architecture type, results, provide textual summaries. Finally, discuss novel techniques data augmentation enhancements, combined advanced architectures, diffusion models, generative hybrid networks, transformers, graph improve analysis.

Language: Английский

Citations

1
Comprehensive assessment of imaging quality of artificial intelligence-assisted compressed sensing-based MR images in routine clinical settings DOI Creative Commons

Adiraju Karthik,

Kamal Aggarwal,

Aakaar Kapoor

et al.

BMC Medical Imaging, Journal Year: 2024, Volume and Issue: 24(1)

Published: Oct. 21, 2024

Conventional MR acceleration techniques, such as compressed sensing, parallel imaging, and half Fourier often face limitations, including noise amplification, reduced signal-to-noise ratio (SNR) increased susceptibility to artifacts, which can compromise image quality, especially in high-speed acquisitions. Artificial intelligence (AI)-assisted sensing (ACS) has emerged a novel approach that combines the conventional techniques with advanced AI algorithms. The objective of this study was examine imaging quality ACS by qualitative quantitative analysis for brain, spine, kidney, liver, knee well compare performance method (non-ACS) imaging. This included 50 subjects. Three radiologists independently assessed images based on artefacts, sharpness, overall diagnostic efficacy. SNR, contrast-to-noise (CNR), edge content (EC), enhancement measure (EME), scanning time were used evaluation. Cohen's kappa correlation coefficient (k) employed radiologists' inter-observer agreement, Mann Whitney U-test comparison between non-ACS ACS. three demonstrated showed superior clinical information than mean k ~ 0.70. acquired statistically higher values (p < 0.05) CNR, EC, EME compared images. Furthermore, study's findings indicated ACS-enabled scan more 50% while maintaining high quality. Integrating technology into routine settings potential speed up acquisition, improve enhance procedures patient throughput.

Language: Английский

Citations

4
Glo-net: A dual task branch based neural network for multi-class glomeruli segmentation DOI
Xiangxue Wang,

Jingkai Zhang,

Yuemei Xu

et al.

Computers in Biology and Medicine, Journal Year: 2025, Volume and Issue: 186, P. 109670 - 109670

Published: Jan. 11, 2025

Language: Английский

Citations

0
Explainable AI associates ECG aging effects with increased cardiovascular risk in a longitudinal population study DOI Creative Commons

Philip Hempel,

Antônio H. Ribeiro, Marcus Vollmer

et al.

npj Digital Medicine, Journal Year: 2025, Volume and Issue: 8(1)

Published: Jan. 13, 2025

Aging affects the 12-lead electrocardiogram (ECG) and correlates with cardiovascular disease (CVD). AI-ECG models estimate aging effects as a novel biomarker but have only been evaluated on single ECGs-without utilizing longitudinal data. We validated an model, originally trained Brazilian data, using German cohort over 20 years of follow-up, demonstrating similar performance (r2 = 0.70) to original study (0.71). Incorporating ECGs revealed stronger association risk, increasing hazard ratio for mortality from 1.43 1.65. Moreover, were associated higher odds ratios atrial fibrillation, heart failure, mortality. Using explainable AI methods that model aligns clinical knowledge by focusing ECG features known reflect aging. Our suggests in can be applied population level identify patients at risk early.

Language: Английский

Citations

0
Age estimation for disorder characterization from pediatric polysomnograms DOI Creative Commons
Sven Festag, Sebastian Herberger, Cord Spreckelsen

et al.

Biomedical Signal Processing and Control, Journal Year: 2025, Volume and Issue: 106, P. 107701 - 107701

Published: Feb. 20, 2025

Language: Английский

Citations

0
ECG-Based Biometric Recognition: A Survey of Methods and Databases DOI Creative Commons

David Meltzer,

David Luengo

Sensors, Journal Year: 2025, Volume and Issue: 25(6), P. 1864 - 1864

Published: March 17, 2025

This work presents a comprehensive and chronologically ordered survey of existing studies data sources on Electrocardiogram (ECG) based biometric recognition systems. is organized in terms the two main goals pursued it: first, description ECG features techniques used literature, including compilation references; second, databases available by referenced studies. The most relevant characteristics are identified, given. To date, no other has presented such complete overview both for ECG-based recognition. Readers interested subject can obtain an understanding state art, easily identifying specific key papers using different criteria, become aware where they test their novel algorithms.

Language: Английский

Citations

0
Profiling electric signals of electrogenic probiotic bacteria using self-attention analysis DOI Creative Commons

Qing Chi,

Jie Tang, Changmian Ji

et al.

Applied Microbiology and Biotechnology, Journal Year: 2025, Volume and Issue: 109(1)

Published: April 22, 2025

Language: Английский

Citations

0