Interpretable discriminant analysis for functional data supported on random non-linear domains DOI Creative Commons
Eardi Lila, Wenbo Zhang,

Swati Rane

et al.

arXiv (Cornell University), Journal Year: 2021, Volume and Issue: unknown

Published: Jan. 1, 2021

We introduce a novel framework for the classification of functional data supported on nonlinear, and possibly random, manifold domains. The motivating application is identification subjects with Alzheimer's disease from their cortical surface geometry associated thickness map. proposed model based upon reformulation problem as regularized multivariate linear regression model. This allows us to adopt direct approach estimation most discriminant direction while controlling its complexity appropriate differential regularization. Our does not require prior covariance structure predictors, which computationally prohibitive in our setting. provide theoretical analysis out-of-sample prediction error explore finite sample performance simulation apply method pooled dataset Disease Neuroimaging Initiative Parkinson's Progression Markers Initiative. Through this application, we identify directions that capture both geometric predictive features are consistent existing neuroscience literature.

Language: Английский

Applications of machine learning and deep learning in SPECT and PET imaging: General overview, challenges and future prospects DOI Creative Commons
C. Jiménez-Mesa, Juan E. Arco, Francisco J. Martínez-Murcia

et al.

Pharmacological Research, Journal Year: 2023, Volume and Issue: 197, P. 106984 - 106984

Published: Nov. 1, 2023

The integration of positron emission tomography (PET) and single-photon computed (SPECT) imaging techniques with machine learning (ML) algorithms, including deep (DL) models, is a promising approach. This enhances the precision efficiency current diagnostic treatment strategies while offering invaluable insights into disease mechanisms. In this comprehensive review, we delve transformative impact ML DL in domain. Firstly, brief analysis provided how these algorithms have evolved which are most widely applied Their different potential applications nuclear then discussed, such as optimization image adquisition or reconstruction, biomarkers identification, multimodal fusion development diagnostic, prognostic, progression evaluation systems. because they able to analyse complex patterns relationships within data, well extracting quantitative objective measures. Furthermore, discuss challenges implementation, data standardization limited sample sizes, explore clinical opportunities future horizons, augmentation explainable AI. Together, factors propelling continuous advancement more robust, transparent, reliable

Language: Английский

Citations

20
The Interplay of Sports and Nutrition in Neurological Health and Recovery DOI Open Access
Vicente Javier Clemente‐Suárez, Laura Redondo-Flórez, Ana Isabel Beltrán-Velasco

et al.

Journal of Clinical Medicine, Journal Year: 2024, Volume and Issue: 13(7), P. 2065 - 2065

Published: April 2, 2024

This comprehensive review explores the dynamic relationship between sports, nutrition, and neurological health. Focusing on recent clinical advancements, it examines how physical activity dietary practices influence prevention, treatment, rehabilitation of various conditions. The highlights role neuroimaging in understanding these interactions, discusses emerging technologies neurotherapeutic interventions, evaluates efficacy sports nutritional strategies enhancing recovery. synthesis current knowledge aims to provide a deeper lifestyle factors can be integrated into improve outcomes.

Language: Английский

Citations

5
Transforming brain research: Neuroimaging breakthroughs driven by AI DOI

Tushita,

Vivek Srivastava, Ravi Kant Singh

et al.

AIP conference proceedings, Journal Year: 2025, Volume and Issue: 3262, P. 020021 - 020021

Published: Jan. 1, 2025

Language: Английский

Citations

0
Statistical Machine-Learning Methods to Model Brain Plasticity DOI
Pablo Robles-Granda, Aron K. Barbey, Oluwasanmi Koyejo

et al.

Oxford University Press eBooks, Journal Year: 2025, Volume and Issue: unknown

Published: March 20, 2025

Abstract This chapter discusses statistical machine-learning (ML) approaches to model brain plasticity, which involves complex changes in the due natural or induced causes. The highlights various advantages that ML models have compared with traditional of plasticity. Since plasticity can be analyzed at levels granularity, this several starting some examples most traditionally studied, is, visual and motor control systems synaptic for memory throughout mammalian neocortex. Then are discussed contexts scales, including main aspects considered multiscale modeling, specific information about neuron level, cortical column, as a result development. Following this, modeling plasticity’s effect on higher-level cognitive functions, specifically those related behavior, cognition, learning, decision making, intelligence, memory. Plasticity when it results from trauma damage is then reviewed. concludes by reviewing open research questions future directions

Language: Английский

Citations

0
HCDPD: A Heterogeneous Causal Framework for Disease Pattern Detection in Medical Imaging DOI
Rongjie Liu, Chengchun Shi, Rui Song

et al.

medRxiv (Cold Spring Harbor Laboratory), Journal Year: 2025, Volume and Issue: unknown

Published: April 16, 2025

Understanding the causal effects of diseases on body organs through medical imaging is crucial for advancing research and improving clinical outcomes. This paper introduces a novel inference framework, Heterogeneous Causal Disease Pattern Detection (HCDPD), designed to map complex pathways from early-stage latent disease patterns their manifestation in as observed later-stage images. HCDPD serves potential outcome framework multivariate responses. It particularly valuable scenarios where patients exhibit significant heterogeneity, while normal controls remain relatively homogeneous. Through application advanced Bayesian techniques, our method effectively estimates both direct indirect within framework. We applied Osteoarthritis Initiative (OAI) dataset, successfully identifying delineating diverse across different patients. capability provides critical insights that can inform early interventions tailor personalized treatment strategies practice.

Language: Английский

Citations

0
RPCA with Log-Schatten Norm and Adaptive Histogram Equalization for Medical Imaging DOI
Habte Tadesse Likassa, Ding‐Geng Chen

International Journal of Statistics in Medical Research, Journal Year: 2025, Volume and Issue: 14, P. 274 - 288

Published: May 3, 2025

Medical imaging, especially cancer and retinal fundus analysis, is often compromised by artifacts heavy noise artifact, which can hinder accurate diagnosis. Existing low-rank sparse component methods, such as RPCA with the conventional nuclear norm, assume uniform singular value weights, may not hold true due to variations in images. We recently developed log-weighted addresses some of these issues but still relies on weight selection, potentially introducing bias. To overcome limitations, we propose a novel method that integrates Log-Schatten Norm (LSN) Adaptive Histogram Equalization (AHE) for medical imaging clinical purposes. The improves penalization structure preservation, while AHE enhances contrast reduces noise. formulated an optimization problem solved using Alternating Direction Method Multipliers (ADMM). Experimental results publicly available image datasets demonstrate our outperforms existing methods enhancing overall quality, making it promising tool applications.

Language: Английский

Citations

0
A general framework of brain region detection and genetic variants selection in imaging genetics DOI

Siqiang Su,

Zhenghao Li, Long Feng

et al.

The Annals of Applied Statistics, Journal Year: 2025, Volume and Issue: 19(2)

Published: May 28, 2025

Language: Английский

Citations

0
NRAAF: A Framework for Comparative Analysis of fMRI Registration Algorithms and Their Impact on Resting-State Neuroimaging Accuracy DOI Creative Commons
Martin Svejda, Nouh Sabri Elmitwally, A. Taufiq Asyhari

et al.

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 47915 - 47941

Published: Jan. 1, 2024

The rapid evolution of neuroimaging techniques underscores the necessity for robust medical image registration algorithms, essential precise analysis resting-state networks. This study introduces a comprehensive modular evaluation framework, designed to assess and compare differences four state-of-the-art algorithms in field: FSL, ANTs, DARTEL, AFNI. Our framework highlights critical importance algorithm selection neuroimaging, addressing unique challenges strengths each presents processing complex brain imaging data. rigorous delves into algorithms' differences, with focus on spatial localisation accuracy fidelity network identification. comparative uncovers distinct advantages limitations inherent algorithm, illuminating how specific characteristics can shape outcomes. For instance, we reveal FSL's robustness handling diverse datasets, ANTs' precision normalization, DARTEL's suitability large-scale studies, AFNI's adaptability functional structural analysis. findings highlight nuanced considerations necessary choosing right data, advocating bespoke approach based requirements study. detailed advances field, guiding researchers towards more informed application, thus aiming improve reliability Presenting clear, overview within our novel addresses needs community paves way future advancements registration.

Language: Английский

Citations

2
Image response regression via deep neural networks DOI
Daiwei Zhang, Lexin Li, Chandra Sripada

et al.

Journal of the Royal Statistical Society Series B (Statistical Methodology), Journal Year: 2023, Volume and Issue: 85(5), P. 1589 - 1614

Published: July 24, 2023

Abstract Delineating associations between images and covariates is a central aim of imaging studies. To tackle this problem, we propose novel non-parametric approach in the framework spatially varying coefficient models, where functions are estimated through deep neural networks. Our method incorporates spatial smoothness, handles subject heterogeneity, provides straightforward interpretations. It also highly flexible accurate, making it ideal for capturing complex association patterns. We establish estimation selection consistency derive asymptotic error bounds. demonstrate method’s advantages intensive simulations analyses two functional magnetic resonance data sets.

Language: Английский

Citations

4
A functional nonlinear mixed effects modeling framework for longitudinal functional responses DOI Creative Commons
Linglong Kong,

Xinchao Luo,

Jinhan Xie

et al.

Electronic Journal of Statistics, Journal Year: 2024, Volume and Issue: 18(1)

Published: Jan. 1, 2024

In this paper, we introduce a functional nonlinear mixed effects modeling framework designed to quantify the random, relationship between individual spatiotemporal trajectories and longitudinal responses. Our proposed accounts for within-individual variability through process. We detail an estimation method determining fixed random effect functions covariance operators establish their asymptotic properties, including uniform consistency weak convergence. also develop global linear hypothesis tests bootstrap-based simultaneous confidence bands functions. To assess finite-sample performance of our method, perform numerical analysis using both simulated real-world datasets. results demonstrate that model class is significantly more flexible effective in detecting compared existing models. apply approach autism research database investigate impact age spatial dynamics on fractional anisotropy along corpus callosum white matter fiber skeleton.

Language: Английский

Citations

1