Enhanced ADHD classification through deep learning and dynamic resting state fMRI analysis DOI Creative Commons
MohammadHadi Firouzi,

Kamran Kazemi,

Maliheh Ahmadi

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Oct. 18, 2024

Attention Deficit Hyperactivity Disorder (ADHD) is characterized by deficits in attention, hyperactivity, and/or impulsivity. Resting-state functional connectivity analysis has emerged as a promising approach for ADHD classification using resting-state magnetic resonance imaging (rs-fMRI), although with limited accuracy. Recent studies have highlighted dynamic changes patterns among children. In this study, we introduce Skip-Vote-Net, novel deep learning-based network designed classifying from typically developing children (TDC) leveraging on rs-fMRI data collected 222 participants included the NYU dataset within ADHD-200 database. Initially, each subject, matrices were constructed overlapping segments Pearson's correlation between mean time series of 116 regions interest defined Automated Anatomical Labeling (AAL) atlas. Skip-Vote-Net was then developed, employing majority voting mechanism to classify ADHD/TDC children, well distinguishing two main subtypes: inattentive subtype (ADHDI) and predominantly combined (ADHDC). The proposed method evaluated across four scenarios: (1) two-class TD balanced data, (2) unbalanced (3) ADHDI ADHDC, (4) three-class ADHDI, Using achieved accuracies 97% ± 1.87 97.7% 2.2 cases, respectively. Furthermore, accuracy discriminating ADHDC reached 99.4% 1.21. Finally, demonstrated an average 98.86% 1.03 collectively. Our findings highlight superior performance over existing methods ADHD, showcasing its potential effective diagnostic tool identifying subtypes

Language: Английский

Predicting functional impairments with lesion‐derived disconnectome mapping: Validation in stroke patients with motor deficits DOI Creative Commons

Maedeh Khalilian,

Martine Roussel,

Olivier Godefroy

et al.

European Journal of Neuroscience, Journal Year: 2024, Volume and Issue: 59(11), P. 3074 - 3092

Published: April 5, 2024

Focal structural damage to white matter tracts can result in functional deficits stroke patients. Traditional voxel-based lesion-symptom mapping is commonly used localize brain structures linked neurological deficits. Emerging evidence suggests that the impact of focal may extend beyond immediate lesion sites. In this study, we present a disconnectome approach based on support vector regression (SVR) identify and pathways associated with For clinical validation, utilized imaging data from 340 patients exhibiting motor A map was initially derived lesions for each patient. Bootstrap sampling then employed balance sample size between minority group right or left those without Subsequently, SVR analysis voxels (p < .005). Our disconnectome-based significantly outperformed alternative approaches identifying major within corticospinal upper-lower limb Bootstrapping increased sensitivity (80%-87%) deficits, minimum 32 235 mm

Language: Английский

Citations

6
Post-Stroke Outcome prediction based on lesion-derived features DOI Creative Commons

Maedeh Khalilian,

Olivier Godefroy,

Martine Roussel

et al.

NeuroImage Clinical, Journal Year: 2025, Volume and Issue: unknown, P. 103747 - 103747

Published: Jan. 1, 2025

Language: Английский

Citations

0
Aperiodic component of EEG power spectrum and cognitive performance are modulated by education in aging DOI Creative Commons
Sonia Montemurro,

Daniel Borek,

Daniele Marinazzo

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: July 2, 2024

Abstract Recent studies have shown a growing interest in the so-called “aperiodic” component of EEG power spectrum, which describes overall trend whole spectrum with linear or exponential function. In field brain aging, this aperiodic is associated both age-related changes and performance on cognitive tasks. This study aims to elucidate potential role education moderating relationship between resting-state features (including component) aging. N = 179 healthy participants “Leipzig Study for Mind–Body-Emotion Interactions” (LEMON) dataset were divided into three groups based age education. Older adults exhibited lower exponent, offset (i.e. measures component), Individual Alpha Peak Frequency (IAPF) as compared younger adults. Moreover, visual attention working memory differently depending education: older high education, higher exponent predicted slower processing speed less capacity, while an opposite was found those low While further investigation needed, shows modulatory aging cognition.

Language: Английский

Citations

3
Enhanced ADHD classification through deep learning and dynamic resting state fMRI analysis DOI Creative Commons
MohammadHadi Firouzi,

Kamran Kazemi,

Maliheh Ahmadi

et al.

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Oct. 18, 2024

Attention Deficit Hyperactivity Disorder (ADHD) is characterized by deficits in attention, hyperactivity, and/or impulsivity. Resting-state functional connectivity analysis has emerged as a promising approach for ADHD classification using resting-state magnetic resonance imaging (rs-fMRI), although with limited accuracy. Recent studies have highlighted dynamic changes patterns among children. In this study, we introduce Skip-Vote-Net, novel deep learning-based network designed classifying from typically developing children (TDC) leveraging on rs-fMRI data collected 222 participants included the NYU dataset within ADHD-200 database. Initially, each subject, matrices were constructed overlapping segments Pearson's correlation between mean time series of 116 regions interest defined Automated Anatomical Labeling (AAL) atlas. Skip-Vote-Net was then developed, employing majority voting mechanism to classify ADHD/TDC children, well distinguishing two main subtypes: inattentive subtype (ADHDI) and predominantly combined (ADHDC). The proposed method evaluated across four scenarios: (1) two-class TD balanced data, (2) unbalanced (3) ADHDI ADHDC, (4) three-class ADHDI, Using achieved accuracies 97% ± 1.87 97.7% 2.2 cases, respectively. Furthermore, accuracy discriminating ADHDC reached 99.4% 1.21. Finally, demonstrated an average 98.86% 1.03 collectively. Our findings highlight superior performance over existing methods ADHD, showcasing its potential effective diagnostic tool identifying subtypes

Language: Английский

Citations

3