Activity¶
85
Commits
17
Active days
1
Contributors
2026-10-13
Last commit
v2.0.1
Latest release
Contributions
Less
More
Show:
2026
Oct 13 · 2 commits
Add an experimental XLE parser and anonymize XLE files
Add an experimental XLE parser and anonymize XLE files
Oct 11 · 7 commits
Files with exactly the same content are the same ECG: they now share one ECG pseudonym, also across runs and patient folders, while each copy is still written. The stats count distinct ECGs from the anonymized file names, so formats without an ECG ID field are counted too.
Each anonymized file is now named <anonymized_patient_id>_<anonymized_ecg_id>.<extension> instead of an edited raw name, since a raw name can hold identity in any form. The raw name stays in the catalog. With --anonymize-patient-folders only the patient folder is renamed, its sub-folders keep their names.
Docs update
Add a duplicate column in CSV for anonymization
Trust the patient folder for the patient ID and count raw and anonymized files from disk
Match the patient folder name inside the patient ID, rename the pseudonym columns and write anonymization stats
Show patient folder name as original patient ID and add anonymization stats
Oct 10 · 5 commits
Add --group-by-patient-folders and replace the patient ID only when trusted
Group patients only by patient folder
Allow anonymizing patient folders without --patients-dir-name
Anonymization: patient folders, single files and UUID ECG pseudonyms
Docs update
Oct 9 · 13 commits
Docs update
ECG anonymization
Rebuild the docs from main when a release is published
Release v1.1.1
Show the detected format and parser in file_format
Add AliveCor Kardia JSON parser
Add AliveCor Kardia JSON parser
Store each wave boundary as its own annotation
Store each wave boundary as its own annotation
Sierra and Mortara parsers stored measurement values both in measurements and in annotations. Values with a measurement field now go only to measurements, all other values stay in annotations.
Docs update
Docs update
The default R-peak detector no longer stops after an artifact, finds no beats on flat signals and no longer counts T waves as beats. - Lead derivation, resampling and normalization give correct values and units on raw and physical leads. - DeepFADE keeps the input amplitude and is marked experimental. - EDAN Holter support is marked experimental. GE MAC 2000 and BeneHeart R12 are removed: they had no specification and no real sample. - DICOM measurements, MFER multi-frame files, GE MUSE units and statements, and several file edge cases are handled correctly. - Batch parsing no longer keeps every record in memory, and long recordings parse and plot with much less memory.
Oct 8 · 5 commits
Stop filling in a sampling rate the file does not give
Warn when the sampling rate is missing from the file
Fix and harden all ECG parsers, report only stored values
Add ChecksumWarning and expose ISHNE checksum validity
Harden ISHNE Holter parser and extend unified schema
Jul 7 · 4 commits
Docs update
Docs update
Docs update
Fix raw-vs-physical lead labeling and DICOM baseline scaling
Jul 6 · 2 commits
Release v1.0.2
ISHNE Holter parser: improve header parsing (pacemaker reading)
May 28 · 3 commits
Add support for EDAN SE Holter ECG (.arc)
Add support for EDAN SE Holter ECG (.arc)
Add support for EDAN SE Holter ECG (.arc)
Mar 20 · 3 commits
Bug fix on tests
Refactor normalization methods for better clarity
Update noramization methods to normalize per lead and per ecg. Add support for 1D, 2D and 3D arrays.
Mar 19 · 8 commits
Support min-max, z-score, and amplitude normalization across multiple leads independently. Accepts list[Lead], 1D array, or 2D array (n_leads x n_samples). Return type matches input type: Lead in → Lead out, numpy in -> numpy out. Add title param to plot_12lead/iplot_12lead, plotly_white template for interactive plots, seaborn-whitegrid for static plots, show_grid defaults to False.
Add list[Lead] and 2D array overloads to normalize functions for per-lead normalization
Switched to seaborn-v0_8-whitegrid — white background with subtle gridlines.
Plots method update
Update plot_12lead method to support title argument. Update style of interactive plots with plotly white
Docs update for 1.0.0 release
Support min-max, z-score, and amplitude normalization across multiple leads independently.
Readme update
Mar 7 · 3 commits
Docs update
Release v0.0.9 — ADC resolution pipeline
Rename sample_rate to sampling_rate across entire codebase
Mar 6 · 4 commits
Add resolution_unit and adc_resolution_unit fields to Lead, update units semantics across all 12 parsers
Auto-detect is_raw from scaling metadata, add adc_resolution field, fix resolution unit conversions.
Auto-detect is_raw from resolution/offset metadata in all parsers, fix plotting and GE MUSE test fixtures
Auto-detect raw ADC signal from scaling metadata in all 12 parsers Parsers no longer hardcode 'is_raw=True' unconditionally. Instead, is_raw is derived from the resolution/offset metadata: when resolution == 1.0 and offset == 0.0 the samples are already in physical units (is_raw=False), otherwise they are raw ADC counts (is_raw=True). This prevents double-scaling when auto_scale=True encounters data that is already in physical units. Updated Lead docstrings to clarify that units describes the target unit when is_raw=True and the actual unit when is_raw=False.
Mar 5
* Automatic scaling of ECG signal from raw ADC when parsing by default (can be disabled with argument), by default parsers will convert if possible to mV (if resolution informations are present) * Methods for converting ECG samples to mV * Docs update
Mar 4 · 4 commits
Gitlab development repository URL fix
Add GitLab install option to getting started docs (development version)
Add ECGRecord __repr__ and plot() method for clean console display and quick plotting
Add FileParser.supported_formats() with format metadata on all 12 parsers
Mar 3 · 6 commits
Simplify multi-lead plots: full signal, configurable grid layout Rewrite plot_leads, plot_12lead, iplot_leads, and iplot_12lead to plot full signals vertically by default with optional rows/cols grid params. plot_12lead and iplot_12lead now assign standard 12-lead names (I, II, III, aVR…V6) to unnamed inputs and no longer crop to a fixed duration or use the paper ECG 4×4 grid layout. Added STANDARD_12LEAD constant and _grid_shape() helper to _core.py. Updated plotting API docs.
Multi-lead numpy array support, Activity page, v0.0.8 - plot_leads, plot_12lead, plot_quality, iplot_leads, iplot_12lead now accept raw numpy arrays with fs=parameter - New LeadsLike type alias for multi-lead inputs - Static plots no longer force matplotlib Agg backend
Docs update
Docs update
Docs update
Docs update
Mar 2 · 14 commits
Docs update
Bug fix in matplotlib static plots - Docs update - Publish to PyPI
Bug fix in matplotlib static plots - Docs update - Publish to PyPI
Bug fix in matplotlib static plots - Docs update
Bug fix in matplotlib static plots - Docs update
Update docs, migrate to sphynx
Update docs
Methods supports now data models and raw numpy arrays
Run tests
Run tests
Run tests
Readme update
Custom domain configured (ecgdatakit.ummisco.fr)
Initial release v0.0.6