Ouija is God‑View for your brain — for the average person. It takes the raw, intimidating signals of neuroscience (EEG, heart rhythm, imaging, body composition) and turns them into a beautiful live 3D brain plus a plain‑English read of what your mind is doing right now and what to do about it. The hard part isn't the hardware; it's making a brain legible to a normal human. That's the product.
Hardware‑agnostic. Ouija ingests from any device that speaks a standard — LSL or BrainFlow for live streams, EDF / CSV for files — and normalizes everything into one model. It is not a bio‑sensing platform tied to one gadget; the gadget is interchangeable.
Validated on real data. The creator (Eros Marcello Iuliano) dog‑foods it with his own recordings — a Neurosity Crown and an Upside Down Labs bio‑sensing kit, plus imaging (SPECT / MRI / fMRI / fNIRS) and a Withings scale — as the validation case that proves the pipeline end‑to‑end. The code and (anonymized) data are open so anyone can follow along, replicate, and point Ouija at their own brain.
The centerpiece is the Brain Atlas — a live 3D brain that populates region‑by‑region as each data modality comes online, reaching a NOMINAL state when everything is feeding. Over the top sits the distiller: it turns the numbers into human sentences ("You're locked in — protect this window for deep work") instead of "F5 band‑power 68%, θ/β 1.4." Deterministic and offline today for robustness; LLM‑upgradable for richer reads.
- Neuroimaging – SPECT scans (rested and active), structural MRIs with shape analysis, guided resting‑state fMRI and fNIRS recordings.
- Biopotentials – EEG data from the Neurosity Crown and the Upside Down Labs bio‑sensing kit, plus EOG, ECG and EMG signals.
- Physiological metrics – weight, body composition, vascular age and heart rhythms from the Withings Full Body Smart Scale.
- Clinical reports – interpretations from neurologists and general practitioners (kept private; only de‑identified derived summaries are ever shared).
- In progress – magnetoencephalography (MEG) and diffusion tensor imaging (DTI) data.
These streams come together to form a multidimensional picture of one mind across time. Raw recordings,
derived features and processing code live in this repository or its linked storage; large datasets are
shared externally with clear documentation, while smaller files are kept here for quick reference. See
bids_dataset/ for the standards‑organized data root.
Consistency is key when measuring the brain. Ouija uses a structured protocol to capture data daily, weekly and monthly. Sessions are tagged and documented to maintain context:
- Baseline setup – Assign a subject ID, synchronize all devices to the same clock and create a session log template. Standardize the recording environment and run signal‑quality checks on EEG headsets and the scale before collecting data.
- Daily routine – Each morning capture weight, body composition, vascular age and ECG via the scale, then record 2 minutes of eyes‑open and 2 minutes of eyes‑closed EEG with the Neurosity Crown. Each evening collect resting EEG using the Upside Down Labs kit, followed by guided breathing and brief notes on stress and energy.
- Weekly blocks – Once a week run a cognitive task battery with the Neurosity Crown and a heart‑breath session with the Upside Down Labs kit to capture how the brain and heart respond to stimuli.
- Monthly integration – Record a narrative note about lifestyle changes, run longer EEG sessions and review body‑composition trends to align with imaging appointments.
- Data hygiene – Label sessions consistently (e.g.
daily_am_rest,weekly_cog), flag poor‑quality recordings, sync device clocks regularly, annotate imaging sessions with context and back up weekly.
The analysis pipeline combines established neuroimaging techniques with custom tooling:
- Structural MRI – Apply statistical shape analysis (e.g. SPHARM‑PDM) and local shape descriptors to quantify morphological differences across sessions.
- Time‑series processing – Filter, segment and feature‑extract EEG, ECG and fNIRS signals; apply dimensionality reduction and machine‑learning models to detect states and patterns.
- Format conversion – Convert per‑device raw exports into standards‑valid BIDS:
EEG → EDF (
mne-bids), fNIRS → SNIRF, MRI/fMRI → NIfTI (dcm2niix/dcm2bids), Withings panels → BIDSphenotype/. Seeconverters/. - Data management – Use Python and Pandas to load, clean and serve data; a FastAPI backend exposes derived summaries to the dashboard.
A local Next.js dashboard (under frontend/) renders the archive as an
interactive "God View": live signal waveforms, a mind‑state estimate, and longitudinal trends. During
development the device feeds are simulated, so the interface runs end‑to‑end without physical hardware
attached.
| Layer | Tools |
|---|---|
| Frontend | Next.js, React, Tailwind CSS, hand‑rolled SVG visualizations |
| Backend | Python with FastAPI, SQLite and Pandas |
| Neuro tooling | MNE‑Python, MNE‑BIDS, pysnirf2, dcm2niix / dcm2bids, bids‑validator |
| Deployment | Vercel for prototypes |
Ouija is an open project. If you care about the brain, code or open knowledge, you are welcome to:
- Fork this repo and run the scripts on your own data.
- Improve analysis routines, visualizations or the cadence protocol.
- Share feedback, open issues or start discussions.
- Contribute neuroscientific insights, machine‑learning methods or full‑stack features.
Please note that this project is experimental and not a diagnostic instrument. It pushes the boundaries of personal neuroscience but is not intended for medical use.
- Query interface – Natural‑language tooling to explore and interpret the archive conversationally.
- Visualization – Richer, more intuitive brain and body dashboards.
- Analysis pipelines – Automate preprocessing and incorporate more advanced shape analysis and predictive models.
- Predictive capabilities – Train models to forecast cognitive or physiological states.
- Community – Expand documentation, publish more datasets and foster a collaborative research network.
- Integration & privacy – Secure APIs for data access and decentralised storage options.
Code is released under the MIT License. Personal neurophysiological data shared here is anonymized and provided under CC0 for research and educational purposes. Use it responsibly, cite the source and respect the privacy of others if you replicate the self‑as‑subject model.
A personal project of Eros Marcello Iuliano.