NeuroPath is grounded in peer-reviewed and accepted research on dementia diagnosis, multimodal neuroimaging, graph learning, and clinically interpretable AI.
Website: neuro-path.ai
| Work | Venue | Short description | Tags | Links |
|---|---|---|---|---|
| MINDSETS Multi-omics Integration with Neuroimaging for Dementia Subtyping and Effective Temporal Study |
Scientific Reports, Nature Portfolio, 2025 | A multi-omics framework that combines longitudinal MRI radiomics with clinical, cognitive, and genetic data to support dementia subtype differentiation across AD, VaD, MCI, and control cohorts. | multi-omics MRI radiomics longitudinal MRI dementia subtyping explainability |
Paper | Code | DOI |
| MAGNET-AD Multitask Spatiotemporal GNN for Interpretable Prediction of PACC and Conversion Time in Preclinical Alzheimer |
MICCAI 2025 | A multitask spatiotemporal graph neural network for preclinical Alzheimer's disease that predicts both cognitive trajectory through PACC and time to AD conversion from longitudinal multimodal data. | spatiotemporal GNN PACC conversion time preclinical AD attention |
Paper | PDF | Code | DOI |
| ClinGRAD Clinically-Guided Genomics and Radiomics Interpretable GNN for Dementia Diagnosis |
MICCAI 2025 | A clinically guided heterogeneous graph model that links patient, brain-region, and gene evidence using MRI radiomics, genomics, DWI priors, and gene co-expression for interpretable dementia diagnosis. | heterogeneous GNN genomics radiomics DWI priors clinical guidance |
Paper | PDF | Code | DOI |
| CONNECT-4 Brain Connectivity-Guided Hyperedge Graph Fusion for Structural MRI to 4D Rest Functional MRI Synthesis |
MICCAI 2026, accepted | A connectivity-guided hyperedge graph fusion framework that synthesizes 4D resting-state fMRI from structural MRI by combining image patches, segmentation patches, ROI features, DWI connectivity, and normative atrophy. | hypergraph fusion 4D rs-fMRI sMRI synthesis DWI connectivity normative atrophy |
Paper coming soon | Code |
flowchart LR
MRI["Structural MRI"] --> MINDSETS["MINDSETS<br/>multi-omics dementia subtyping"]
MRI --> CONNECT["CONNECT-4<br/>sMRI to 4D rs-fMRI synthesis"]
MINDSETS --> CLINGRAD["ClinGRAD<br/>clinically guided graph diagnosis"]
MINDSETS --> MAGNET["MAGNET-AD<br/>preclinical trajectory modeling"]
CLINGRAD --> NEUROPATH["NeuroPath<br/>research-backed dementia AI"]
MAGNET --> NEUROPATH
CONNECT --> NEUROPATH
| Layer | What it adds to NeuroPath |
|---|---|
| Differential diagnosis | MINDSETS establishes a multi-omics foundation for distinguishing dementia subtypes from MRI-derived and tabular evidence. |
| Clinical interpretability | ClinGRAD connects diagnosis to brain regions, genes, pathways, and clinical priors through graph reasoning. |
| Progression modeling | MAGNET-AD adds longitudinal risk estimation through cognitive trajectory and conversion-time prediction. |
| Functional synthesis | CONNECT-4 extends structural MRI workflows toward functional connectivity analysis when resting-state fMRI is unavailable. |