SMAS: Structural MRI-Based AD Score using Bayesian VAE DOI Creative Commons
Aditya Nemali, José Bernal, Renat Yakupov

et al.

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

Published: Nov. 7, 2024

Abstract This study introduces the Structural MRI-based Alzheimer’s Disease Score (SMAS), a novel index intended to quantify (AD)-related morphometric patterns using deep learning Bayesian-supervised Variational Autoencoder (Bayesian-SVAE). SMAS was constructed baseline structural MRI data from DELCODE and evaluated longitudinally in two independent cohorts: DEL-CODE (n=415) ADNI (n=190). Our findings indicate that has strong associations with cognitive performance (DELCODE: r=-0.83; ADNI: r=-0.62), age (DEL-CODE: r=0.50; r=0.28), hippocampal volume r=-0.44; r=-0.66), total grey matter r=-0.42; r=-0.47), suggesting its potential as biomarker for AD-related brain atrophy. Moreover, our longitudinal studies suggest may be useful early identification tracking of AD. The model demonstrated significant predictive accuracy distinguishing cognitively healthy individuals those AD AUC=0.971 at baseline, 0.833 36 months; AUC=0.817 improving 0.903 24 months). Notably, over 36-month period, outperformed existing measures such SPARE-AD volume. Relevance map analysis revealed morphological changes key regions—including hippocampus, posterior cingulate cortex, precuneus, lateral parietal cortex—highlighting is sensitive interpretable atrophy, suitable detection monitoring disease progression.

Language: Английский

Data–model Fusion Methods and Applications toward Smart Manufacturing and Digital Engineering DOI Creative Commons
Fei Tao, Yilin Li, Yupeng Wei

et al.

Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 1, 2025

Language: Английский

Citations

1
Artificial intelligence and the diagnosis of oral cavity cancer and oral potentially malignant disorders from clinical photographs: a narrative review DOI Creative Commons

Payam Mirfendereski,

Grace Y. Li,

Alexander T. Pearson

et al.

Frontiers in Oral Health, Journal Year: 2025, Volume and Issue: 6

Published: March 10, 2025

Oral cavity cancer is associated with high morbidity and mortality, particularly advanced stage diagnosis. cancer, typically squamous cell carcinoma (OSCC), often preceded by oral potentially malignant disorders (OPMDs), which comprise eleven variable risks for transformation. While OPMDs are clinical diagnoses, conventional exam followed biopsy histopathological analysis the gold standard diagnosis of OSCC. There vast heterogeneity in presentation OPMDs, possible visual similarities to early-stage OSCC or even various benign mucosal abnormalities. The diagnostic challenge OSCC/OPMDs compounded non-specialist primary care setting. has been significant research interest technology assist OSCC/OPMDs. Artificial intelligence (AI), enables machine performance human tasks, already shown promise several domains medical diagnostics. Computer vision, field AI dedicated data, over past decade applied photographs Various methodological concerns limitations may be encountered literature on OSCC/OPMD image analysis. This narrative review delineates current landscape photograph navigates limitations, issues, workflow implications this field, providing context future considerations.

Language: Английский

Citations

1
Decoding uncertainty for clinical decision-making DOI Creative Commons
Krasimira Tsaneva‐Atanasova,

Giulia Pederzanil,

Marianna Laviola

et al.

Philosophical Transactions of the Royal Society A Mathematical Physical and Engineering Sciences, Journal Year: 2025, Volume and Issue: 383(2292)

Published: March 13, 2025

In this opinion piece, we examine the pivotal role that uncertainty quantification (UQ) plays in informing clinical decision-making processes. We explore challenges associated with healthcare data and potential barriers to widespread adoption of UQ methodologies. doing so, highlight how these techniques can improve precision reliability medical evaluations. delve into crucial understanding managing uncertainties present (such as measurement error), diagnostic tools treatment outcomes. discuss such impact emphasize importance systematically analysing them. Our goal is demonstrate effectively addressing decoding significantly enhance accuracy robustness decisions, ultimately leading better patient outcomes more informed practices. This article part theme issue ‘Uncertainty for biological systems (Part 1)’.

Language: Английский

Citations

1
Has multimodal learning delivered universal intelligence in healthcare? A comprehensive survey DOI Creative Commons
Qika Lin, Y. C. Zhu, Mei Xin

et al.

Information Fusion, Journal Year: 2024, Volume and Issue: unknown, P. 102795 - 102795

Published: Nov. 1, 2024

Language: Английский

Citations

8
Deep evidential fusion with uncertainty quantification and reliability learning for multimodal medical image segmentation DOI
Ling Huang, Su Ruan, Pierre Decazes

et al.

Information Fusion, Journal Year: 2024, Volume and Issue: 113, P. 102648 - 102648

Published: Aug. 23, 2024

Language: Английский

Citations

5
Non-parametric Bayesian deep learning approach for whole-body low-dose PET reconstruction and uncertainty assessment DOI Creative Commons
Maya Fichmann Levital, Samah Khawaled, John A. Kennedy

et al.

Medical & Biological Engineering & Computing, Journal Year: 2025, Volume and Issue: unknown

Published: Jan. 23, 2025

Positron emission tomography (PET) imaging plays a pivotal role in oncology for the early detection of metastatic tumors and response to therapy assessment due its high sensitivity compared anatomical modalities. The balance between image quality radiation exposure is critical, as reducing administered dose results lower signal-to-noise ratio (SNR) information loss, which may significantly affect clinical diagnosis. Deep learning (DL) algorithms have recently made significant progress low-dose (LD) PET reconstruction. Nevertheless, successful application requires thorough evaluation uncertainty ensure informed judgment. We propose NPB-LDPET, DL-based non-parametric Bayesian framework LD reconstruction assessment. Our utilizes an Adam optimizer with stochastic gradient Langevin dynamics (SGLD) sample from underlying posterior distribution. employed Ultra-low-dose Challenge dataset assess our framework's performance relative Monte Carlo dropout benchmark. evaluated global accuracy utilizing SSIM, PSNR, NRMSE, local lesion conspicuity using mean absolute error (MAE) contrast, relevance maps employing correlation measures reduction factor (DRF). NPB-LDPET method exhibits superior various DRFs (paired t-test, p<0.0001 , N=10, 631). Moreover, we demonstrate 21% MAE (573.54 vs. 723.70, paired N=28) 8.3% improvement contrast (2.077 1.916, N=28). Furthermore, stronger predicted 95th percentile score DRF ( r2=0.9174 r2=0.6144 proposed has potential improve decision-making by providing more accurate informative while exposure.

Language: Английский

Citations

0
Towards reliable land cover mapping under domain shift: An overview and comprehensive comparative study on uncertainty estimation DOI
Chao Ji,

Hong Tang

Earth-Science Reviews, Journal Year: 2025, Volume and Issue: unknown, P. 105070 - 105070

Published: Feb. 1, 2025

Language: Английский

Citations

0
Uncertainty-based fatigue life of vehicle subframe via improved bootstrap method under load extrapolation DOI Creative Commons

W.Q. Li,

Xintian Liu,

Weihao Su

et al.

Proceedings of the Institution of Mechanical Engineers Part D Journal of Automobile Engineering, Journal Year: 2025, Volume and Issue: unknown

Published: Feb. 27, 2025

Fatigue failures in vehicle subframes are a critical challenge due to complex and unpredictable loads. Traditional methods often fail capture the uncertainty load conditions, resulting unreliable fatigue life predictions. This study introduces an improved bootstrap method address these uncertainties. Real-world testing data were used construct spectra with Generalized Pareto Distribution model, enabling accurate prediction of rare but impactful events. The rain-flow counting was perform frequency statistics on signals. obtained S-N curve corrected based Haibach theory. process provided distribution parameters mean amplitude. then estimated using modified Miner’s theory, which achieved significant improvements accuracy reliability. improves can be applied product design improvement mechanical engineering related fields.

Language: Английский

Citations

0
Deep learning-based uncertainty quantification for quality assurance in hepatobiliary imaging-based techniques DOI Open Access
Yashbir Singh, Jesper B. Andersen, Quincy A. Hathaway

et al.

Oncotarget, Journal Year: 2025, Volume and Issue: 16(1), P. 249 - 255

Published: Jan. 20, 2025

Recent advances in deep learning models have transformed medical imaging analysis, particularly radiology. This editorial outlines how uncertainty quantification through embedding-based approaches enhances diagnostic accuracy and reliability hepatobiliary imaging, with a specific focus on oncological conditions early detection of precancerous lesions. We explore modern architectures like the Anisotropic Hybrid Network (AHUNet), which leverages both 2D 3D volumetric data innovative convolutional approaches. consider implications for quality assurance radiological practice discuss recent clinical applications.

Language: Английский

Citations

0
Evidential time-to-event prediction with calibrated uncertainty quantification DOI Creative Commons
Ling Huang, Yucheng Xing, Swapnil Mishra

et al.

International Journal of Approximate Reasoning, Journal Year: 2025, Volume and Issue: unknown, P. 109403 - 109403

Published: March 1, 2025

Language: Английский

Citations

0