NEURAL: A Neonatal EEG Feature Set in Matlab

Matlab code to generate a set of quantitative features from multichannel EEGrecordings. Features include amplitude measures, spectral measures, and basic connectivitymeasures (across hemisphere's only). Also, for preterm EEG (assuming gestational age < 32weeks), will generate features from bursts annotations (e.g. maximum inter-burstinterval). Burst annotations require a separate package, also available ongithub.

Releases archived at Zenodo:

Full details of the methods are in:

JM O’Toole and GB Boylan (2017). NEURAL: quantitative features for newborn EEG using Matlab. ArXiv e-prints, arXiv:1704.05694

which is available at https://arxiv.org/abs/1704.05694.

Requirements | Use | Quantitativefeatures | Files | Test computersetup | Licence | References |Contact

Requirements

Matlab (R2013 or newer,Mathworks) with the signalprocessing toolbox and statistics toolbox. Not tested with Octave but should work with minortweaking.

Use

Set paths in Matlab, or do so using the load_curdir function:

>> load_curdir;

As an example, generate simulated EEG and calculate relative spectral power, standarddeviation of range-EEG, and brain symmetry index:

% generate EEG-like data (coloured Gaussian noise)
	data_st=gen_test_EEGdata(5*60,64,1);

	% define feature set (or can define in neural_parameters.m):
	feature_set={'spectral_relative_power','rEEG_SD', 'connectivity_BSI'};
	
	% estimate features:
	feat_st=generate_all_features(data_st,[],feature_set);

See the demos/ directory for further examples. All parameters are set the fileneural_parameters.m.

Quantitative features

The feature set contains amplitude, spectral, connectivity, and burst annotation features.Amplitude features include range-EEG (D. O’ Reilly et al., 2012;see references), a clearly-defined alternative to amplitude-integrated EEG(aEEG). All features are generated for four different frequency bands (typically 0.5–4,4–7, 7–13, and 13–30 Hz), with some exceptions. The following table describes the featuresin more detail:

FB: features generated for each frequency band (FB)

Files

Some Matlab files (.m files) have a description and an example in the header. To read thisheader, type help <filename.m> in Matlab. Directory structure is as follows:

├── amplitude_features/                  # amplitude features
├── spectral_features/                   # spectral features
├── connectivity_features/               # hemisphere connectivity features
├── range_EEG/                           # range-EEG (similar to aEEG)
├── IBI_features/                        # features from the burst annotations 
├── preprocessing/                       # loads EEG from EDF files (including artefact removal)
└── utils/                               # misc. functions

with some files of interest:

├── neural_parameters.m                  # all parameters defined here
├── all_features_list.m                  # complete list of functions (do not edit)
└── generate_all_features.m              # main function: generates feature set on EEG

Test computer setup

• hardware: Intel Core i7-8700K @ 4.5GHz; 32GB memory.
• operating system: Ubuntu GNU/Linux x86_64 distribution (19.10) with Linux kernel5.3.0-24-generic
• software: Matlab (R2019a)

Licence

Copyright (c) 2016, John M. O' Toole, University College Cork
All rights reserved.

Redistribution and use in source and binary forms, with or without modification,
are permitted provided that the following conditions are met:

  Redistributions of source code must retain the above copyright notice, this
  list of conditions and the following disclaimer.

  Redistributions in binary form must reproduce the above copyright notice, this
  list of conditions and the following disclaimer in the documentation and/or
  other materials provided with the distribution.

  Neither the name of the University College Cork nor the names of its
  contributors may be used to endorse or promote products derived from
  this software without specific prior written permission.

THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND
ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR CONTRIBUTORS BE LIABLE FOR
ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
(INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON
ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

References

1. JM O’Toole and GB Boylan (2017). NEURAL: quantitative features for newborn EEG usingMatlab. ArXiv e-prints, arXiv:1704.05694.

2. D O’Reilly, MA Navakatikyan, M Filip, D Greene, & LJ Van Marter (2012). Peak-to-peakamplitude in neonatal brain monitoring of premature infants. Clinical Neurophysiology,123(11):2139–2153.

3. MJAM van Putten (2007). The revised brain symmetry index. Clinical Neurophysiology,118(11):2362–2367.

4. T Higuchi (1988). Approach to an irregular time series on the basis of the fractal theory,Physica D: Nonlinear Phenomena, 31:277–283.

5. MJ Katz (1988). Fractals and the analysis of waveforms. Computers in Biology andMedicine, 18(3):145–156.

6. AV Oppenheim, RW Schafer. Discrete-Time Signal Processing. Prentice-Hall, EnglewoodCliffs, NJ 07458, 1999.

7. JM O’ Toole, GB Boylan, S Vanhatalo, NJ Stevenson (2016). Estimating functional brainmaturity in very and extremely preterm neonates using automated analysis of theelectroencephalogram. Clinical Neurophysiology, 127(8):2910–2918

8. JM O’ Toole, GB Boylan, RO Lloyd, RM Goulding, S Vanhatalo, NJ Stevenson(2017). Detecting Bursts in the EEG of Very and Extremely Premature Infants using aMulti-Feature Approach. Medical Engineering andPhysics, vol. 45, pp. 42-50, 2017.DOI:10.1016/j.medengphy.2017.04.003

9. JM O'Toole and GB Geraldine (2019). Quantitative Preterm EEG Analysis: The Need forCaution in Using Modern Data Science Techniques. Frontiers in Pediatrics 7, 174DOI:10.3389/fped.2019.00174

Contact

John M. O'Toole

Neonatal Brain Research Group,
INFANT Research Centre,
Department of Paediatrics and Child Health,
Room 2.19 UCC Academic Paediatric Unit, Cork University Hospital,
University College Cork,
Ireland

• email: jotoole AT ucc _dot ie