seiz_eeg.dataset

EEG Data class with common data retrieval

class seiz_eeg.dataset.EEGDataset(clips_df: DataFrame[ClipsDF], *, signal_transform: Callable[[ndarray[Any, dtype[float64]]], ndarray[Any, dtype[_ScalarType_co]] | Any] | None = None, label_transform: Callable[[int], Any] | None = None, prefetch: bool = False, diff_channels: bool = False)

Dataset of EEG clips with seizure labels

Parameters:

• clips_df (DataFrame[ClipsDF]) – Pandas dataframe of EEG clips annotations

• signal_transform (Callable[[NDArray[float]], NDArray | Any], optional) – Function to transform signals before they are returned. Defaults to None.

• label_transform (Callable[[int], Any], optional) – Function to transform labels before they are returned. Defaults to None.

• prefetch (bool, optional) – Wether to prefetch all clips. Defaults to False.

• diff_channels (bool, optional) – Wether to subtract pairwise channels. Defaults to False.

clip_lenght

Lenght of each clip in seconds.

Type:

float

s_rate

Sampling rate of the clips.

Type:

int

output_shape

Shape of the output tensors.

Type:

Tuple[tuple, tuple]

class seiz_eeg.dataset.EEGFileDataset(clips_df: DataFrame[ClipsDF], *, signal_transform: Callable[[ndarray[Any, dtype[float64]]], ndarray[Any, dtype[_ScalarType_co]] | Any] | None = None, label_transform: Callable[[int], Any] | None = None, diff_channels: bool = False)

Extension of EEGDataset which returns a tensor of all clips from the same file.

seiz_eeg.dataset.to_arrays(data: EEGDataset, pbar=False) → Tuple[ndarray[Any, dtype[float64]], ndarray[Any, dtype[int64]]]

Load all signals from data into a tensor.

Parameters:

• data (EEGDataset) – Clips dataset. All clips must have the same lenght to be stacked.

• pbar (bool, optional) – Wether toshow a progress bar while loading. Defaults to False.

Returns:

Signals and labels

Return type:

Tuple[NDArray[float], NDArray[int]]