Skip to content

Latest commit

 

History

History
161 lines (133 loc) · 5.18 KB

CHANGELOG.md

File metadata and controls

161 lines (133 loc) · 5.18 KB

Change Log

All notable changes to this project will be documented in this file. Project started 2016-04-01 but this log started 2017-03-16, after version 0.2.2. More details can be found in the git logs.

unreleased

Changed

Removed

Fixed

Added

[0.4.4] - 26-08-2020

Changed

  • Returns Table of feature values (per epoch and per channel) from function generate_all_features if input argument return_feat_epoch=true

Removed

Fixed

  • Bug fix for connectivity_features; was: if channel labels (cell of strings) are empty then would assume (incorrectly) only 2 channels and would use index [1,2] for left,right hemisphere; now: if no channel labels throw warning

Added

  • Check in spectral_features: is L_window longer than signal (and if so throw warning)?

[0.4.3] - 2020-04-22

Changed

  • changed behaviour for IBI_burst_prc and IBI_burst_number features: now removing any NaNs before calculating this metric

Removed

Fixed

  • bug in rand_phase.m when taking the inverse-FFT (random signal will time-reversed but with same initial x[0] value); won't affect the coherence values.
  • for rectangular window in filter_zerophase.m don't divide by window length
  • for IBI features fixed removal of minimal interval of 16 samples and set to 1/4 seconds instead
  • avoid error if maximum/median IBI duration is zero
  • bug fix for spectral_diff feature: extra zero was included in spectral difference measure before taking the median

Added

[0.4.2] - 2019-04-29

Changed

  • connectivity analysis: remove channel pairs that have common electrode, e.g. 'C3-Cz' and 'Cz-C4'

Removed

Fixed

  • tested with Matlab R2019a
  • fixed syntax problem for fir1 function for R2019a

Added

[0.4.1] - 2019-03-19

Changed

  • minor tweak to the overlap-and-add analysis when generating features on an epoch of EEG; now include as much of the signal as possible and pad with NaNs when necessary.

Removed

Fixed

  • tested with Matlab R2018b;
  • bug when downsampling: always using the resample function even when not necessary;
  • bug in PSD power feature if length-epoch < (2 * PSD-window-length);
  • bug when length-epoch = signal-length, was summing over frequency bands.

Added

[0.4.0] - 2019-01-09

Changed

  • Syntax for specifying the zero-level threshold of the coherence function in neural_parameters.m. NB: this will break compatibility with previous versions.

Removed

  • 'robust-PSD' for coherence function (as does not satisfy the Schwartz inequality).

Fixed

  • Coherence function for surrogate analysis, which assumed that auto-PSDs Pxx and Pyy would not change when randomising the phase of the Fourier transform for the surrogate signals. This holds only when PSD is a periodogram. Assumption removed and now Pxx and Pyy generated for each iteration.

Added

  • Analytic zero-coherence threshold (Halliday et al. 1995) for coherence estimates. Estimates coherence using the Bartlett PSD.

[0.3.4] - 2018-09-18

Changed

  • Higuchi method for fractal dimension can now incorporate NaNs

Removed

Fixed

  • tested with Matlab R2018a

Added

[0.3.3] - 2017-11-16

Changed

Removed

Fixed

  • tested with Matlab R2017a

Added

  • licence file
  • DOI links from zenodo

[0.3.2] - 2017-06-16

Changed

Removed

Fixed

  • reduced order of high-pass butterworth filter (with cut-off of 0.1Hz; see change in v0.3.1) from 5 to 2 in remove_artefacts.m function

Added

  • name of .mat file as input argument to 'resample_savemat.m'
  • option in filter_butterworth_withnans.m to have separate filter orders for high and low-pass components of a bandpass filter
  • file to plot rEEG (plot_rEEG.m)

[0.3.1] - 2017-04-07

Changed

  • bandpass filter changed from 0.5-40 Hz to 0.1-40 Hz before removing large-amplitude artefacts

Removed

  • bandpass filtering (0.5-40 Hz) on EEG prior to removing artefacts (constant values or sudden jumps in amplitude)

Fixed

  • avoid failing if only 1 channel for connectivity features (return NaN instead)

Added

[0.3.0] - 2017-03-29

Changed

  • absolute and relative spectral power ('spectral_power' and 'spectral_relative_power') now using periodogram (rectangular window) instead of Welch's power-spectral-density (PSD) estimate.
  • all spectral estimates (PSD, robust-PSD, and periodogram) now used FFT length equal to window length (previously was in 2^n form, and at least >=256), i.e. no zero-padding.

Removed

  • PSD option for absolute and relative spectral power removed (replaced with periodogram).

Fixed

  • coherence function broken
  • split of frequency bands for spectral and connectivity features when length of PSD is odd

Added

  • documentation to file headers
  • this CHANGELOG.md!
  • option to select PSD estimate for spectral features: either Welch PSD, robust-PSD (modified Welch PSD), or periodogram
  • coherence function can use either PSD or robust-PSD

[0.2.2] - 2017-03-09

Added

  • surrogate data approach with coherence function
  • NaN option to 'FILTER_REPLACE_ARTEFACTS'
  • missing 'get_window.m' and 'make_odd.m' function
  • now using semantic versioning (see http://semver.org/)

Fixed

  • scaling factor for robust-PSD