Automated labelling of Movement-Related Cortical Potentials using Segmented Regression

Usman Rashid, Imran Khan Niazi, Mads Jochumsen, Laurens R Krol, Nada Signal, Denise Taylor

Research output: Contribution to journalJournal articleResearchpeer-review

22 Downloads (Pure)

Abstract

The movement-related cortical potential (MRCP) is a brain signal related to planning and execution of motor tasks. From an MRCP, three notable features can be identified: the early Bereitschaftspotential (BP1), the late Bereitschaftspotential (BP2), and the negative peak (PN). These features have been used in past studies to quantify neurophysiological changes in response to motor training. Currently, either manual labelling or a priori specification of time points is used to extract these features. The limitation of these methods is the inability to fully model the features. This study proposes segmented regression along with a local peak method for automated labelling of the features. The proposed method derives the onsets, amplitudes at onsets and slopes of BP1, BP2 along with time and amplitude of the negative peak (PN) in a typical average MRCP. To choose the most suitable regression technique bounded segmented regression, a change point method and multivariate adaptive regression splines were evaluated using root mean square error on a dataset of 6000 simulated MRCPs. The best performing regression technique combined with the local peak method was then applied to a smaller set of 123 simulated MRCPs. Error in onsets of BP1, BP2, and time of PN were compared with errors in manual labelling by an expert. The performance of the proposed method was also evaluated on an experimental dataset of MRCPs derived from electroencephalography (EEG) recorded across two sessions from 22 healthy participants during a lower limb task. Bland-Altman plots were used to evaluate the absolute reliability of the proposed method. On experimental data, the proposed method was also compared to manual labelling by an expert. Bounded segmented regression produced the smallest error on the simulation data. For the experimental data, our proposed method did not exhibit statistically significant bias in any of the modelled features. Furthermore, its performance was comparable to manual labelling by experts. We conclude that the proposed method be used to automatically obtain robust estimates for the MRCP features with known measurement error.

Original languageEnglish
JournalI E E E Transactions on Neural Systems and Rehabilitation Engineering
ISSN1534-4320
DOIs
Publication statusE-pub ahead of print - 7 May 2019

Fingerprint

Labeling
Contingent Negative Variation
Electroencephalography
Measurement errors
Mean square error
Splines
Brain
Specifications
Planning
Lower Extremity
Healthy Volunteers

Cite this

@article{dc493cb14c7f4557ba215adad327a957,
title = "Automated labelling of Movement-Related Cortical Potentials using Segmented Regression",
abstract = "The movement-related cortical potential (MRCP) is a brain signal related to planning and execution of motor tasks. From an MRCP, three notable features can be identified: the early Bereitschaftspotential (BP1), the late Bereitschaftspotential (BP2), and the negative peak (PN). These features have been used in past studies to quantify neurophysiological changes in response to motor training. Currently, either manual labelling or a priori specification of time points is used to extract these features. The limitation of these methods is the inability to fully model the features. This study proposes segmented regression along with a local peak method for automated labelling of the features. The proposed method derives the onsets, amplitudes at onsets and slopes of BP1, BP2 along with time and amplitude of the negative peak (PN) in a typical average MRCP. To choose the most suitable regression technique bounded segmented regression, a change point method and multivariate adaptive regression splines were evaluated using root mean square error on a dataset of 6000 simulated MRCPs. The best performing regression technique combined with the local peak method was then applied to a smaller set of 123 simulated MRCPs. Error in onsets of BP1, BP2, and time of PN were compared with errors in manual labelling by an expert. The performance of the proposed method was also evaluated on an experimental dataset of MRCPs derived from electroencephalography (EEG) recorded across two sessions from 22 healthy participants during a lower limb task. Bland-Altman plots were used to evaluate the absolute reliability of the proposed method. On experimental data, the proposed method was also compared to manual labelling by an expert. Bounded segmented regression produced the smallest error on the simulation data. For the experimental data, our proposed method did not exhibit statistically significant bias in any of the modelled features. Furthermore, its performance was comparable to manual labelling by experts. We conclude that the proposed method be used to automatically obtain robust estimates for the MRCP features with known measurement error.",
author = "Usman Rashid and Niazi, {Imran Khan} and Mads Jochumsen and Krol, {Laurens R} and Nada Signal and Denise Taylor",
year = "2019",
month = "5",
day = "7",
doi = "10.1109/TNSRE.2019.2913880",
language = "English",
journal = "I E E E Transactions on Neural Systems and Rehabilitation Engineering",
issn = "1534-4320",
publisher = "IEEE",

}

Automated labelling of Movement-Related Cortical Potentials using Segmented Regression. / Rashid, Usman; Niazi, Imran Khan; Jochumsen, Mads; Krol, Laurens R; Signal, Nada; Taylor, Denise.

In: I E E E Transactions on Neural Systems and Rehabilitation Engineering, 07.05.2019.

Research output: Contribution to journalJournal articleResearchpeer-review

TY - JOUR

T1 - Automated labelling of Movement-Related Cortical Potentials using Segmented Regression

AU - Rashid, Usman

AU - Niazi, Imran Khan

AU - Jochumsen, Mads

AU - Krol, Laurens R

AU - Signal, Nada

AU - Taylor, Denise

PY - 2019/5/7

Y1 - 2019/5/7

N2 - The movement-related cortical potential (MRCP) is a brain signal related to planning and execution of motor tasks. From an MRCP, three notable features can be identified: the early Bereitschaftspotential (BP1), the late Bereitschaftspotential (BP2), and the negative peak (PN). These features have been used in past studies to quantify neurophysiological changes in response to motor training. Currently, either manual labelling or a priori specification of time points is used to extract these features. The limitation of these methods is the inability to fully model the features. This study proposes segmented regression along with a local peak method for automated labelling of the features. The proposed method derives the onsets, amplitudes at onsets and slopes of BP1, BP2 along with time and amplitude of the negative peak (PN) in a typical average MRCP. To choose the most suitable regression technique bounded segmented regression, a change point method and multivariate adaptive regression splines were evaluated using root mean square error on a dataset of 6000 simulated MRCPs. The best performing regression technique combined with the local peak method was then applied to a smaller set of 123 simulated MRCPs. Error in onsets of BP1, BP2, and time of PN were compared with errors in manual labelling by an expert. The performance of the proposed method was also evaluated on an experimental dataset of MRCPs derived from electroencephalography (EEG) recorded across two sessions from 22 healthy participants during a lower limb task. Bland-Altman plots were used to evaluate the absolute reliability of the proposed method. On experimental data, the proposed method was also compared to manual labelling by an expert. Bounded segmented regression produced the smallest error on the simulation data. For the experimental data, our proposed method did not exhibit statistically significant bias in any of the modelled features. Furthermore, its performance was comparable to manual labelling by experts. We conclude that the proposed method be used to automatically obtain robust estimates for the MRCP features with known measurement error.

AB - The movement-related cortical potential (MRCP) is a brain signal related to planning and execution of motor tasks. From an MRCP, three notable features can be identified: the early Bereitschaftspotential (BP1), the late Bereitschaftspotential (BP2), and the negative peak (PN). These features have been used in past studies to quantify neurophysiological changes in response to motor training. Currently, either manual labelling or a priori specification of time points is used to extract these features. The limitation of these methods is the inability to fully model the features. This study proposes segmented regression along with a local peak method for automated labelling of the features. The proposed method derives the onsets, amplitudes at onsets and slopes of BP1, BP2 along with time and amplitude of the negative peak (PN) in a typical average MRCP. To choose the most suitable regression technique bounded segmented regression, a change point method and multivariate adaptive regression splines were evaluated using root mean square error on a dataset of 6000 simulated MRCPs. The best performing regression technique combined with the local peak method was then applied to a smaller set of 123 simulated MRCPs. Error in onsets of BP1, BP2, and time of PN were compared with errors in manual labelling by an expert. The performance of the proposed method was also evaluated on an experimental dataset of MRCPs derived from electroencephalography (EEG) recorded across two sessions from 22 healthy participants during a lower limb task. Bland-Altman plots were used to evaluate the absolute reliability of the proposed method. On experimental data, the proposed method was also compared to manual labelling by an expert. Bounded segmented regression produced the smallest error on the simulation data. For the experimental data, our proposed method did not exhibit statistically significant bias in any of the modelled features. Furthermore, its performance was comparable to manual labelling by experts. We conclude that the proposed method be used to automatically obtain robust estimates for the MRCP features with known measurement error.

U2 - 10.1109/TNSRE.2019.2913880

DO - 10.1109/TNSRE.2019.2913880

M3 - Journal article

JO - I E E E Transactions on Neural Systems and Rehabilitation Engineering

JF - I E E E Transactions on Neural Systems and Rehabilitation Engineering

SN - 1534-4320

ER -