systole.detection.ppg_peaks#

systole.detection.ppg_peaks(signal: Union[List, ndarray, Series], sfreq: int, new_sfreq: int = 1000, method: str = 'rolling_average', clipping: bool = True, clipping_thresholds: Union[Tuple, List, str] = 'auto', clean_nan: bool = False, verbose: bool = False, detector_kws: Dict = {}) → Tuple[ndarray, ndarray][source]#

Systolic peak detection for PPG signals.

Two methods are available: - an adaptation of the rolling average + standard deviation approach described in [R57c6d7365c14-1]. - The Multi-scale peak and trough detection algorithm (MSPTD) [R57c6d7365c14-2].

Before peaks detection, nans are interpolated (optional, Fale by default) the signal is resampled to the new sampling frequency (1000 Hz by default) and clipping artefacts are corrected using cubic spline interpolation (optional, True by default).

Note

This function will resample the signal to 1000 Hz by default.

Parameters

signal

The raw signal recorded from the pulse oximeter time series.

sfreq

The sampling frequency (Hz).

new_sfreq

If resample is True, the new sampling frequency (Hz). Defaults to 1000.

method

The systolic peaks detection algorithm to use, can be “rolling_average” [R57c6d7365c14-1] (default) or “msptd” [R57c6d7365c14-2].

clipping

If True, will apply the clipping artefact correction described in [R57c6d7365c14-1]. Defaults to True.

clipping_thresholds

The values of the minumum and maximum clipping thresholds. Can be a float or None. If None, no correction is applied. If “auto” is provided, will use

systole.utils.find_clipping() to find the values. Defaults to “auto”.: This parameter is only relevant if cliping is True.

clean_nan

If True, will interpolate NaNs values if any before any other operation. Defaults to False.

verbose

Control function verbosity. Defaults to False (do not print processing steps).

detector_kws

Additional keyword arguments that will be passed to the detector function.

Returns

resampled_signal: Signal resampled to the new_sfreq frequency.
peaks: Boolean array of systolic peaks detection.

Raises

ValueError: If clipping_thresholds is not a tuple, a list or “auto”. If method is not a valid method name.

References

R57c6d7365c14-1(1,2,3): van Gent, P., Farah, H., van Nes, N. and van Arem, B., 2019. Analysing Noisy Driver Physiology Real-Time Using Off-the-Shelf Sensors: Heart Rate Analysis Software from the Taking the Fast Lane Project. Journal of Open Research Software, 7(1), p.32. DOI: http://doi.org/10.5334/jors.241
R57c6d7365c14-2(1,2): S. M. Bishop and A. Ercole, ‘Multi-scale peak and trough detection optimised for periodic and quasi-periodic neuroscience data,’ in Intracranial Pressure and Neuromonitoring XVI. Acta Neurochirurgica Supplement, T. Heldt, Ed. Springer, 2018, vol. 126, pp. 189-195. <https://doi.org/10.1007/978-3-319-65798-1_39>

Examples

>>> from systole import import_ppg
>>> from systole.detection import ppg_peaks
>>> ppg = import_ppg().ppg.to_numpy()  # Import PPG signal

Using the rolling average method (default)#

>>> signal, peaks = ppg_peaks(signal=ppg, method="rolling_average")
>>> print(f'{sum(peaks)} peaks detected.')
378 peaks detected.

Using the Multi-scale peak and trough detection algorithm#

>>> signal, peaks = ppg_peaks(signal=ppg, method="msptd")
>>> print(f'{sum(peaks)} peaks detected.')
378 peaks detected.