Dataset workspace
Open any dataset from the study home page to reach /areas/neurology/studies/{id}/datasets/{dsid} — the main workspace for one recording or matrix.
Connectivity matrix datasets
Graph metrics panel
Run Compute graph metrics to calculate BCT-equivalent network measures:
- Global: mean clustering, characteristic path length, global efficiency, modularity, small-worldness (sigma)
- Per-node: degree, strength, clustering, betweenness, local efficiency, module assignment
Results appear in metric cards and a sortable per-node table.
Visualizations
When graph metrics complete:
- Heatmap — connectivity matrix with optional threshold slider
- Network graph — force-directed layout; nodes colored by module
- 3D brain view — region positions from atlas coordinates when labels match a supported atlas
Interpret
Launch Interpret after graph metrics complete for a narrative summary citing your BCT values.
EEG datasets
Preprocessing
Preprocess EEG applies MNE-Python preprocessing:
- Bandpass and notch filtering
- Bad channel detection and interpolation
- Average or custom re-reference
- Optional ICA artifact rejection
Review channel count, duration, and quality summary when complete.
Features
Compute EEG features (requires completed preprocess):
- Band power: delta, theta, alpha, beta, gamma
- Spectral entropy
- Connectivity: PLV, coherence, wPLI (configurable methods)
Connectivity matrices from features can feed Compute graph metrics.
Advanced analysis
When preprocessing is complete:
- Microstates — polarity-insensitive k-means segmentation with coverage and transition statistics
- Time-frequency — spectrogram and band-power dynamics
- Source localization — simplified source-space summary (research exploratory)
Each advanced pipeline runs asynchronously with status polling on the page.
fMRI datasets
BIDS validation
For BIDS ZIP uploads, run Validate BIDS to check dataset structure and surface missing fields before FC analysis.
Functional connectivity
Compute functional connectivity extracts ROI timeseries via atlas parcellation (AAL, Schaefer 100/200/400), applies confound regression and bandpass filtering, and produces a Pearson correlation matrix with Fisher z transformation.
Dynamic FC
Compute dynamic FC (requires FC or preprocessed input) runs sliding-window correlation and k-means state identification.
Graph metrics on FC
Run Compute graph metrics on the resulting connectivity matrix to obtain network topology measures.
Running pipelines
- Click the pipeline action (for example Compute graph metrics)
- The job enters
running— status updates on the page - On completion, panels and visualizations refresh automatically
- Failed runs show error messages in the job result
Job status
The page refreshes status automatically while jobs are in progress — no need to reload.