CPSC2018

class torch_ecg.databases.CPSC2018(db_dir: str | bytes | PathLike | None = None, working_dir: str | bytes | PathLike | None = None, verbose: int = 1, **kwargs: Any)

Bases: CPSCDataBase

The China Physiological Signal Challenge 2018: Automatic identification of the rhythm/morphology abnormalities in 12-lead ECGs.

ABOUT

1. training set contains 6,877 (female: 3178; male: 3699) 12 leads ECG recordings lasting from 6 s to just 60 s.

2. ECG recordings were sampled as 500 Hz.

3. the training data can be downloaded using links in [1], but the link in [2] is recommended. File structure will be assumed to follow [2].

4. the training data are in the channel first format.

5. types of abnormal rhythm/morphology + normal in the training set:

   No.	name	abbr.	number of records
   0	Normal	N	918
   1	Atrial fibrillation	AF	1098
   2	First-degree atrioventricular block	I-AVB	704
   3	Left bundle brunch block	LBBB	207
   4	Right bundle brunch block	RBBB	1695
   5	Premature atrial contraction	PAC	556
   6	Premature ventricular contraction	PVC	672
   7	ST-segment depression	STD	825
   8	ST-segment elevated	STE	202

6. ordering of the leads in the data of all the records are

   ["I", "II", "III", "aVR", "aVL", "aVF", "V1", "V2", "V3", "V4", "V5", "V6"]
   

7. meanings in the .hea files: to write

8. knowledge about the abnormal rhythms: ref. get_disease_knowledge().

9. Challenge official website [1], see also [2].

Note

1. Ages of records A0608, A1549, A1876, A2299, A5990 are “NaN”.

2. CINC2020 (ref. [2]) released totally 3453 unused training data of CPSC2018, whose filenames start with “Q”. These file names are not “continuous”. The last record is “Q3581”.

Usage

1. ECG arrythmia detection

References

[1] (1,2)

http://2018.icbeb.org/

[2] (1,2,3,4)

https://moody-challenge.physionet.org/2020/

Citation

10.1166/jmihi.2018.2442

Parameters:

• db_dir (path-like, optional) – Storage path of the database. If not specified, data will be fetched from Physionet.

• working_dir (path-like, optional) – Working directory, to store intermediate files and log files.

• verbose (int, default 1) – Level of logging verbosity.

• kwargs (dict, optional) – Auxilliary key word arguments.

property database_info: DataBaseInfo

The DataBaseInfo object of the database.

download() → None

Download the database from self.url.

get_labels(rec: str | int, ann_format: str = 'n') → List[str]

Load labels (diagnoses or arrhythmias) of a record.

Parameters:

• rec (str or int) – Record name or index of the record in all_records.

• ann_format (str, default "n") –

  Format of labels, one of the following (case insensitive):

  • ”a”, abbreviations

  • ”f”, full names

  • ”n”, numeric codes

Returns:

labels – The list of labels.

Return type:

List[str]

get_subject_id(rec: int | str) → int

Attach a unique subject ID for the record.

Parameters:

rec (str or int) – Record name or index of the record in all_records.

Returns:

Subject ID associated with the record.

Return type:

int

get_subject_info(rec: int | str, items: List[str] | None = None) → dict

Get subject information (e.g sex, age, etc.).

Parameters:

• rec (int or str) – Record name or index of the record in all_records.

• items (List[str], optional) – Items of the subject information (e.g. sex, age, etc.).

Returns:

subject_info – The subject information.

Return type:

dict

load_ann(rec: str | int, ann_format: str = 'n') → List[str]

Load labels (diagnoses or arrhythmias) of a record.

Parameters:

• rec (str or int) – Record name or index of the record in all_records.

• ann_format (str, default "n") –

  Format of labels, one of the following (case insensitive):

  • ”a”, abbreviations

  • ”f”, full names

  • ”n”, numeric codes

Returns:

labels – The list of labels.

Return type:

List[str]

load_data(rec: int | str, leads: str | int | Sequence[int | str] | None = None, data_format='channel_first', units: str = 'mV', return_fs: bool = False) → ndarray | Tuple[ndarray, Real]

Load the ECG data of a record.

Parameters:

• rec (int or str) – Record name or index of the record in all_records.

• leads (str or int or Sequence[str] or Sequence[int], optional) – The leads to load, None or “all” for all leads.

• data_format (str, default "channel_first") – Format of the ECG data, “channel_last” or “channel_first” (original)

• units (str, default "mV") – Units of the output signal, can also be “μV” (with an alias “uV”), case insensitive.

• return_fs (bool, default False) – Whether to return the sampling frequency of the output signal.

Returns:

• data (numpy.ndarray) – The loaded ECG data.

• data_fs (numbers.Real, optional) – Sampling frequency of the output signal. Returned if return_fs is True.

plot(rec: int | str, ticks_granularity: int = 0, leads: str | List[str] | None = None, **kwargs: Any) → None

Plot the ECG data of a record.

Parameters:

• rec (int or str) – Record name or index of the record in all_records.

• ticks_granularity (int, default 0) – Granularity to plot axis ticks, the higher the more ticks. 0 (no ticks) –> 1 (major ticks) –> 2 (major + minor ticks)

• leads (str or List[str], optional) – The leads to plot

• kwargs (dict, optional) – Auxilliary key word arguments to pass to matplotlib.pyplot.subplots().

property url: List[str]

URL(s) for downloading the database.

Added in version 0.0.4.