ECG morphology phenotyping
Move beyond rhythm and rate into waveform shape — delineate P/Q/R/S/T landmarks on a recording, derive PR, QRS, QT/QTc, ST-level, and T-wave metrics per beat, and verify the placement visually in the waveform explorer.
Research question
What are the ECG interval and morphology characteristics (PR, QRS duration, QT/QTc, ST level) of this recording, and are the fiducial landmarks placed reliably enough to trust the summaries?
Who this is for
- Cardiovascular researchers studying conduction and repolarization (QT/QTc, QRS) rather than autonomic variability
- Drug-safety and cardiac-safety teams screening interval changes across recordings
- Labs that already detect R-peaks and want clinically familiar ECG phenotype metrics
Data requirements
| Data | Required | Purpose |
|---|---|---|
| ECG waveform | Yes | Cleaning, R-peak detection, and delineation |
| WFDB pair, CSV, or EDF | Supported formats | Automatic format detection on upload |
| Completed preprocess run | Recommended | Reuses detected R-peaks so morphology covers the same beats |
Workflow
Create study → Upload ECG → Preprocess → Delineate ECG → Explorer (landmark overlay) → Interpret (optional)
Step 1 — Upload and preprocess
Upload an ECG dataset, then go to Analyze → Signal Explorer and run Preprocess ECG on the Overview tab. This detects R-peaks and scores signal quality.
Step 2 — Delineate ECG
Click Delineate ECG in the morphology panel. The pipeline reuses the preprocess R-peaks, locates P/Q/R/S/T fiducials, and computes per-beat and recording-level morphology:
| Metric | Description |
|---|---|
| PR interval | P-wave onset to QRS onset |
| QRS duration | QRS onset to offset |
| QT / QTc | QRS onset to T-offset, rate-corrected (Bazett and Fridericia) |
| ST level | Amplitude at J-point + 60 ms relative to the isoelectric baseline |
| T amplitude | T-peak amplitude relative to baseline |
The panel reports median and IQR per interval plus the percentage of beats successfully delineated.
Step 3 — Verify in the explorer
Open the Explorer tab and enable the Landmarks toggle. Zoom to a few beats to see color-coded P/Q/R/S/T markers on the trace — use this to confirm delineation quality before trusting the summaries.
Step 4 — Interpret (optional)
Run Interpret to get a descriptive narrative that cites the computed morphology metrics. QTc is flagged as assuming sinus rhythm, and a caveat is added when few beats are delineated.
Expected outputs
- Per-beat fiducial landmarks and morphology intervals
- Recording-level median/IQR for PR, QRS, QT, QTc, ST, and T-wave metrics
- Percentage of beats delineated as a quality gate
- Landmark overlay in the waveform explorer
- AI narrative citing morphology metrics (when interpretation is run)
Example insight
Median QTc (Bazett) is prolonged relative to typical resting norms with a normal QRS duration, and 96% of beats were delineated — the repolarization signal is reliable, but QTc should be read in the context of the recording's rhythm stability.