dspk.synapse.features

Waveform features: shape (morphology), Haar wavelet coefficients, and the conduction velocity of an HD-EMG motor unit. Principal components are ops.linalg.PCA.

conduction_velocity	Muscle-fibre conduction velocity of an HD-EMG motor unit, from its template.
haar_wavelet	Multi-level Haar wavelet transform of a signal.
morphology	Shape features of single-channel waveforms.

Functions

conduction_velocity

conduction_velocity(
    template,
    rows: int,
    cols: int,
    pitch_um: float,
    sample_rate: float,
) -> dict

Muscle-fibre conduction velocity of an HD-EMG motor unit, from its template.

Parameters

• template : array_like
(rows * cols, samples) on a row-major grid.
• rows : int
  The grid.
• cols : int
  The grid.
• pitch_um : float
  Electrode spacing, in µm.
• sample_rate : float
  Sample rate in Hz.

Returns

• dict — velocity_m_per_s, mean_delay_ms, correlation.

haar_wavelet

haar_wavelet(signal, levels: int = 4) -> np.ndarray

Multi-level Haar wavelet transform of a signal.

Parameters

• signal : array_like
  1-D.
• levels : int, default 4
  Decomposition levels.

Returns

• np.ndarray — The coefficients concatenated, approximation first.

morphology

morphology(waveforms) -> dict

Shape features of single-channel waveforms.

Parameters

• waveforms : array_like
(spikes, samples) in µV, one channel.

Returns

• dict of np.ndarray — Per waveform: trough_uv, trough_to_peak_samples, half_width_samples, repolarization_slope.