Review the recording
Browse raw synthetic EEG, change the display montage, open full-screen review and inspect a linked spectrogram. Display changes preserve the underlying synthetic signal.

CORTILUME / EEG SOFTWARE · IN DEVELOPMENT
Cortilume is the EEG software being developed for Sentinel: a workspace for recording, monitoring, review, analysis and reporting. Its proposed clinical workflow can be explored in DOWS.
In development · Workflow demo available
Cortilume is in development. Its DOWS demonstration uses fictional episodes and synthetic signals. Live hardware, clinical video, AI detection and hospital integrations are not implemented clinical services.
The aim is to keep the original EEG, recording quality, relevant observations and the clinician’s decisions together. At the bedside, the first question is what needs attention. For the reviewer, it is which evidence supports an interpretation. The interface should make both easy to find.
Sentinel is the EEG hardware programme. Cortilume is its EEG software. DOWS is the onboarding environment in which people can explore and practise the proposed workflow. Daughter remains the separate proposal for shared records, permissions and follow-up across Phanes programmes.
We are exploring scalp heatmaps, estimated cortical source maps and regional trends for Sentinel’s planned 64- and 128-channel configurations. Local processing is the intended direction, with raw EEG retained for review.
The protected HD-EEG explorer demonstrates playback, signal-quality checks and electrode-position checks using illustrative data. No inverse solver or trained model runs in this exercise; personalised anatomy is a workflow example, not a loaded patient MRI.
Explore HD-EEG in DOWS →The protected DOWS station is a working demonstration using fictional episodes and synthetic signals. These interactions are implemented; their clinical effectiveness has not been established.
Browse raw synthetic EEG, change the display montage, open full-screen review and inspect a linked spectrogram. Display changes preserve the underlying synthetic signal.
Select a stretch of recorded EEG by dragging, or enter its boundaries. Save a label and note, then return to the same interval from the event record. Flags are annotations; they do not diagnose a seizure, raise an alarm or count as completed review.
Practise accepting a review, documenting findings, confirming communication, recording a clinical decision and handing over outstanding work. Bedside assessment runs in parallel with specialist review.
Explore a separate multi-bed overview and export fixed-template handover, EEG report and patient-history PDFs, plus a structured JSON workflow record. Saved interval flags are included.
A linked bedside illustration demonstrates the shared recording cursor and video-unavailable states. It is not patient footage. The optional review guide retrieves recorded demo state through fixed rules; it is not a diagnostic AI or a trained EEG model.
CLINICAL WORKSPACE
Live Sentinel acquisition, original-signal preservation, timestamped video, multi-bed monitoring and remote specialist review. Real hardware synchronisation, hospital connectivity and service reliability still need to be engineered and tested.
ANALYSIS SUPPORT
Signal-quality checks, quantitative EEG trends and candidate-event detection, with uncertainty, missing coverage and model provenance visible. A clinician’s interpretation remains distinct from an automated candidate. No EEG model is deployed in the demo.
RECORDING CONTINUITY
Local storage, battery-supported acquisition, integrity checks and reconciliation after a connection interruption are engineering goals. The demo illustrates parts of the workflow; it does not test battery runtime, electrical protection or real data-loss performance.
RESEARCH ACCESS
Planned access to authorised signal data, metadata and processing history through documented exports and Python or MATLAB workflows. EDF/BIDS-compatible exchange is a design direction, not an available export claim. HDF5 storage and BIDS organisation solve different problems and are not interchangeable formats.
Our initial clinical design focus is adult neuro-ICU monitoring. Neonatal monitoring, ambulatory recording, stroke-related analysis, traumatic brain injury applications and anatomical source localisation require their own requirements, datasets and validation. They are research directions, not demonstrated Cortilume diagnostic capabilities.
Power optimisation must preserve the required acquisition. Patient inactivity alone would not justify lowering the recording quality. Proposed artefact processing must preserve access to the original signal and show where evidence is uncertain or missing.
The intended clinical system needs explicit access permissions, audit history, secure transfer, retention controls and clear ownership of each record. Research access and model training need separate governance; participant corrections are not automatically fed into a learning model.
DOWS must use fictional training details. Its current storage and feedback behaviour is described in our data-handling page. A demonstration is not evidence of production security, regulatory approval or clinical performance.
We welcome clinicians, EEG practitioners and researchers who can test whether setup, review, communication and handover are understandable in practice. The useful feedback is specific: where information was hard to find, a responsibility was unclear, or a step added work.