LongRecordingAnalyzer

class neurodent.core.LongRecordingAnalyzer(longrecording: LongRecordingOrganizer, fragment_len_s=10, apply_notch_filter=True)

Bases: object

Class for analyzing LongRecordings by breaking them into smaller time windows (fragments).

Parameters:

• longrecording (LongRecordingOrganizer) – The LongRecordingOrganizer instance to analyze.

• fragment_len_s (int, optional) – Length of adjacent windows in seconds. Defaults to 10.

• apply_notch_filter (bool, optional) – Whether to apply notch filter (constants.LINE_FREQ) to loaded fragments. Defaults to True.

Variables:

• LongRecording (LongRecordingOrganizer) – The LongRecordingOrganizer instance being analyzed.

• fragment_len_s (int) – Length of each analysis fragment/window in seconds.

• n_fragments (int) – Total number of fragments in the recording.

• channel_names (list[str]) – List of channel names.

• n_channels (int) – Number of channels.

• mult_to_uV (float) – Multiplier to convert raw data to microvolts.

• f_s (int) – Sampling frequency in Hz.

• apply_notch_filter (bool) – Whether notch filtering is applied to fragments.

__init__(longrecording: LongRecordingOrganizer, fragment_len_s=10, apply_notch_filter=True) → None

get_fragment_rec(index) → BaseRecording

Get window at index as a spikeinterface recording object

Parameters:

index (int) – Index of time window

Returns:

spikeinterface recording object with optional notch filtering applied

Return type:

si.BaseRecording

get_fragment_np(index, recobj=None) → ndarray

Get window at index as a numpy array object

Parameters:

• index (int) – Index of time window

• recobj (si.BaseRecording, optional) – If not None, uses this recording object to get the numpy array. Defaults to None.

Returns:

Numpy array with dimensions (N, M), N = number of samples, M = number of channels. Values in uV

Return type:

np.ndarray

get_fragment_mne(index, recobj=None) → ndarray

Get window at index as a numpy array object, formatted for ease of use with MNE functions

Parameters:

• index (int) – Index of time window

• recobj (si.BaseRecording, optional) – If not None, uses this recording object to get the numpy array. Defaults to None.

Returns:

Numpy array with dimensions (1, M, N), M = number of channels, N = number of samples. 1st dimension corresponds

to number of epochs, which there is only 1 in a window. Values in uV

Return type:

np.ndarray

get_file_end(index, **kwargs)

compute_rms(index, **kwargs)

Compute average root mean square amplitude

Parameters:

index (int) – Index of time window

Returns:

np.ndarray with shape (1, M), M = number of channels

Return type:

result

compute_logrms(index, **kwargs)

Compute the log of the root mean square amplitude

compute_ampvar(index, **kwargs)

Compute average amplitude variance

Parameters:

index (int) – Index of time window

Returns:

np.ndarray with shape (1, M), M = number of channels

Return type:

result

compute_logampvar(index, **kwargs)

Compute the log of the amplitude variance

compute_psd(index, welch_bin_t=1, notch_filter=True, multitaper=False, **kwargs)

Compute PSD (power spectral density)

Parameters:

• index (int) – Index of time window

• welch_bin_t (float, optional) – Length of time bins to use in Welch’s method, in seconds. Defaults to 1.

• notch_filter (bool, optional) – If True, applies notch filter at line frequency. Defaults to True.

• multitaper (bool, optional) – If True, uses multitaper method instead of Welch’s method. Defaults to False.

Returns:

Array of sample frequencies psd (np.ndarray): Array of PSD values at sample frequencies. (X, M), X = number of sample frequencies, M = number of channels. If sample window length is too short, PSD is interpolated

Return type:

f (np.ndarray)

compute_psdband(index, welch_bin_t=1, notch_filter=True, bands: list[tuple[float, float]] = {'alpha': (8, 13), 'beta': (13, 25), 'delta': (1, 4), 'gamma': (25, 40), 'theta': (4, 8)}, multitaper=False, **kwargs)

Compute power spectral density of the signal for each frequency band.

Parameters:

• index (int) – Index of time window

• welch_bin_t (float, optional) – Length of time bins to use in Welch’s method, in seconds. Defaults to 1.

• notch_filter (bool, optional) – If True, applies notch filter at line frequency. Defaults to True.

• bands (list[tuple[float, float]], optional) – List of frequency bands to compute PSD for. Defaults to constants.FREQ_BANDS.

• multitaper (bool, optional) – If True, uses multitaper method instead of Welch’s method. Defaults to False.

Returns:

Dictionary mapping band names to PSD values for each channel

Return type:

dict

compute_logpsdband(index, welch_bin_t=1, notch_filter=True, bands: list[tuple[float, float]] = {'alpha': (8, 13), 'beta': (13, 25), 'delta': (1, 4), 'gamma': (25, 40), 'theta': (4, 8)}, multitaper=False, **kwargs)

Compute the log of the power spectral density of the signal for each frequency band.

compute_psdtotal(index, welch_bin_t=1, notch_filter=True, band: tuple[float, float] = (1, 40), multitaper=False, **kwargs)

Compute total power over PSD (power spectral density) plot within a specified frequency band

Parameters:

• index (int) – Index of time window

• welch_bin_t (float, optional) – Length of time bins to use in Welch’s method, in seconds. Defaults to 1.

• notch_filter (bool, optional) – If True, applies notch filter at line frequency. Defaults to True.

• band (tuple[float, float], optional) – Frequency band to calculate over. Defaults to constants.FREQ_BAND_TOTAL.

• multitaper (bool, optional) – If True, uses multitaper method instead of Welch’s method. Defaults to False.

Returns:

(M,) long array, M = number of channels. Each value corresponds to sum total of PSD in that band at that channel

Return type:

psdtotal (np.ndarray)

compute_logpsdtotal(index, welch_bin_t=1, notch_filter=True, band: tuple[float, float] = (1, 40), multitaper=False, **kwargs)

Compute the log of the total power over PSD (power spectral density) plot within a specified frequency band

compute_psdfrac(index, welch_bin_t=1, notch_filter=True, bands: list[tuple[float, float]] = {'alpha': (8, 13), 'beta': (13, 25), 'delta': (1, 4), 'gamma': (25, 40), 'theta': (4, 8)}, total_band: tuple[float, float] = (1, 40), multitaper=False, **kwargs)

Compute the power spectral density in each band as a fraction of the total power.

compute_logpsdfrac(index, welch_bin_t=1, notch_filter=True, bands: list[tuple[float, float]] = {'alpha': (8, 13), 'beta': (13, 25), 'delta': (1, 4), 'gamma': (25, 40), 'theta': (4, 8)}, total_band: tuple[float, float] = (1, 40), multitaper=False, **kwargs)

Compute the log of the power spectral density in each band as a fraction of the total power.

compute_psdslope(index, welch_bin_t=1, notch_filter=True, band: tuple[float, float] = (1, 40), multitaper=False, **kwargs)

Compute the slope of the power spectral density of the signal.

Parameters:

• index (int) – Index of time window

• welch_bin_t (float, optional) – Length of time bins to use in Welch’s method, in seconds. Defaults to 1.

• notch_filter (bool, optional) – If True, applies notch filter at line frequency. Defaults to True.

• band (tuple[float, float], optional) – Frequency band to calculate over. Defaults to constants.FREQ_BAND_TOTAL.

• multitaper (bool, optional) – If True, uses multitaper method instead of Welch’s method. Defaults to False.

Returns:

Array of shape (M,2) where M is number of channels. Each row contains [slope, intercept] of log-log fit.

Return type:

np.ndarray

convert_idx_to_timebound(index: int) → tuple[float, float]

Convert fragment index to timebound (start time, end time)

Parameters:

index (int) – Fragment index

Returns:

Timebound in seconds

Return type:

tuple[float, float]

compute_cohere(index, freq_res: float = 1, mode: Literal['cwt_morlet', 'multitaper'] = 'multitaper', geomspace: bool = False, cwt_n_cycles_max: float = 7.0, mt_bandwidth: float = 4.0, downsamp_q: int = 4, epsilon: float = 0.01, **kwargs) → ndarray

compute_zcohere(index, z_epsilon: float = 1e-06, **kwargs) → ndarray

Compute the Fisher z-transformed coherence of the signal.

Parameters:

• index (int) – Index of time window

• z_epsilon (float) – Small value to prevent arctanh(1) = inf. Values are clipped to [-1+z_epsilon, 1-z_epsilon]

• **kwargs – Additional arguments passed to compute_zcohere

compute_imcoh(index, **kwargs) → ndarray

compute_zimcoh(index, z_epsilon: float = 1e-06, **kwargs) → ndarray

compute_pcorr(index, lower_triag=False, **kwargs) → ndarray

compute_zpcorr(index, z_epsilon: float = 1e-06, **kwargs) → ndarray

Compute the Fisher z-transformed Pearson correlation coefficient of the signal.

Parameters:

• index (int) – Index of time window

• z_epsilon (float) – Small value to prevent arctanh(1) = inf. Values are clipped to [-1+z_epsilon, 1-z_epsilon]

• **kwargs – Additional arguments passed to compute_zpcorr

compute_nspike(index, **kwargs)

compute_lognspike(index, **kwargs)