Delphos — User Guide (Graphical Interface)¶

Delphos automatically detects spikes and high-frequency oscillations in an EEG/MEG/SEEG recording, and adds a marker for each detected event.
1. Launching Delphos¶
Menu Process → Signal → Delphos 1.0.1.
Delphos analyzes the channels displayed in the current view: - If channels are selected in the signal view, only those are analyzed. - Otherwise, all channels of the current montage are analyzed. - A time selection made beforehand in the view (click-drag on the trace) can also be used to restrict the analysis to a precise portion of the recording.
2. Choosing the data portion to analyze¶
If the file already contains markers, a first window opens to precisely define which portions to work on: - Check "Click to avoid chunks of data" to exclude certain portions: pick a marker from the drop-down list then "Add to list" to add it to the list of sections to ignore ("Clear list" to empty it). - Check "Click to use chunks of data" to instead keep ONLY the portions covered by certain markers: same principle ("Add to list" / "Clear"). - If neither box is checked, Delphos analyzes the entire recording.
If the file has no markers at all, this step is skipped automatically and Delphos analyzes the whole file.
3. Setting the detection parameters¶
A second window ("Delphos") then opens with three groups of settings:
Wavelet settings - Number of voices (default 12) - Vanish moment (default 20)
Detection settings - Threshold (default 40) - Osc. width threshold (default 1.4) - Osc. freq. spread threshold (default 12) - Spike width threshold (default 1.3) - Spike freq. spread threshold (default 11)
Select events to detect - Spike checkbox (checked by default) - Oscillation checkbox (unchecked by default) — once checked, two fields unlock to specify the frequency band to analyze (Hz), automatically capped at a quarter of the file's sampling rate.
These default values suit most recordings; adjusting them is mainly useful if standard detection misses visibly present events, or conversely detects too many.
Click "Detect" to launch the analysis (or "Close" to cancel).
4. Result¶
Delphos runs in the background — a notification appears in the Windows notification area once the
analysis is finished ("Process Delphos has finished"). Detected events appear directly as markers
on the analyzed channels, in the signal view, and are automatically saved to the .mrk file
associated with the recording.
Reading the produced markers¶
| Marker | Meaning | Color |
|---|---|---|
Spike |
Detected spike | black |
Infra slow |
Oscillation < 1 Hz | very dark blue |
Delta |
1 – 3.5 Hz | blue |
Theta |
3.5 – 7.4 Hz | light blue |
Alpha |
7.4 – 12.4 Hz | cyan |
Beta |
12.4 – 24 Hz | green |
Gamma |
24 – 80 Hz | amber |
Ripple |
80 – 250 Hz | orange |
Fast Ripple |
250 – 500 Hz | red |
Very Fast Osc |
> 500 Hz | magenta |
Only the bands covered by the Oscillation range chosen in step 3 can appear. Each marker is positioned on its originating channel; its value (visible by hovering / in the marker's properties) indicates the detected frequency in Hz.
Tips¶
- To re-analyze only a few suspicious channels, select them in the view before relaunching Delphos rather than rerunning it on everything.
- The parameters entered in the "Delphos" window are not remembered from one run to the next — they reset to their default values every time.