Stavros I. Dimitriadis

This study evaluates bivariate and multivariate connectivity estimators in MEG data to improve virtual source connectivity analysis for resting state recordings, finding that multivariate approaches outperform bivariate methods in classification and repeatability.

The paper evaluates 576 functional connectomics pipelines to identify those that minimize spurious discrepancies and motion confounds while being sensitive to inter-subject and experimental effects.

This study examines the repeatability and graph construction dependence of structural brain network properties, finding subject-specific Laplacian eigenvalues and high identification accuracy with specific graph schemes.

The paper assesses the repeatability of multi-frequency multilayer brain network topologies across different researcher choices in preprocessing and analysis.

This study evaluates how different data-processing pipelines affect the reproducibility of brain network topologies across repeated scans, identifying reliable pipeline combinations.

The paper assesses the repeatability of community detection algorithms and graph construction schemes in structural brain networks using Human Connectome Project data, finding optimal pairings for reliable community and hub identification.

This study identifies a structural brain network hub in the right paracentral lobule associated with dementia risk factors in asymptomatic individuals, without affecting other graph theoretical measures.

This paper develops a high-order dynamic functional connectivity graph (HO-IDFCG) from EEG data to identify schizophrenia biomarkers, achieving 98% classification accuracy using machine learning.

The paper develops a method using MEG data to identify mild cognitive impairment by analyzing dynamic functional connectivity microstates and achieves 85% classification accuracy.

This study evaluates 20 graph construction schemes for structural brain networks, finding that scheme choice significantly affects reproducibility and statistical power.

This paper uses MEG and dynamic connectivity analysis to identify altered brain network microstates in mTBI patients compared to controls.

The paper introduces novel complexity and flexibility indexes to analyze brain activity and connectivity across multiple neuroimaging modalities and conditions, showing their ability to distinguish schizophrenia and detect anesthesia effects.

This study evaluates the reliability of static and dynamic functional connectivity metrics from resting-state MEG data, finding that dynamic metrics and chronnectomic parameters are more reliable than static node-wise metrics.

The paper analyzes resting-state neuromagnetic signals in children with and without reading difficulties using symbolic mutual information to identify network abnormalities associated with reading ability.

This paper introduces a single-sensor automatic sleep-stage classification method using cross-frequency coupling estimates, achieving high accuracy and outperforming previous studies.

The paper uses a Random Forest-based ensemble strategy to classify four groups of patients using MRI features from the ADNI database, achieving 61.9% accuracy in a blind validation set.

This paper introduces a data-driven topological filtering method using orthogonal minimal spanning trees to analyze resting-state MEG connectivity, showing improved classification of brain injury and developmental groups.

This study examines changes in brain connectivity patterns in mild traumatic brain injury using resting-state MEG data, finding altered coupling modes and high classification accuracy between mTBI and control groups.