| DOI | Resolve DOI: https://doi.org/10.1109/IEMTRONICS51293.2020.9216365 |
|---|
| Author | Search for: Hajra, Sujoy Ghosh; Search for: Gopinath, Shishir; Search for: Liu, Careesa C.; Search for: Pawlowski, Gabriela; Search for: Fickling, Shaun D.; Search for: Song, Xiaowei; Search for: D'Arcy, Ryan C.N. |
|---|
| Format | Text, Article |
|---|
| Conference | 2020 IEEE International IOT, Electronics and Mechatronics Conference (IEMTRONICS), September 9-12, 2020, Vancouver, BC, Canada |
|---|
| Subject | biomedical signal processing; neurotechnology; electroencephalography (EEG); neural signal processing; empirical mode decomposition (EMD); artifact removal; electrodes; noise reduction; arrays; time series analysis; correlation; transforms |
|---|
| Abstract | Background: Brain function assessments based on event-related potentials (ERPs) derived from electroencephalography (EEG) are increasingly being conducted in realistic out-of-the-laboratory settings for clinical and non-clinical uses. For rapid testing and practical limitations, such applications require the use of low-density electrode arrays. A major impediment to their use in these applications is the lack of denoising techniques capable of removing artefactual contamination and isolating the ERPs features of interest within low-density arrays. Methods: A novel denoising technique combining empirical mode decomposition (EMD) with template matching procedure is developed and applied to individual-channel data, and the results of this new approach are compared to the results of a conventional (independent component analysis) denoising approach. Both whole-epoch morphological comparisons and specific ERP feature amplitude comparisons were undertaken at the group and individual level for a variety of ERPs indexing sensory (N100), attention (P300) and language processing (N400) using data from 31 healthy adults. Results: The new denoising technique successfully enables the capture of ERPs ranging from low-level sensation to attention to language processing (all p<; 0.05). Intra-class correlation analysis reveals high degree of similarity in the time series waveforms derived from the new and the conventional denoising approaches for all ERPs (highest r=0.89, all p<; 0.001). Analysis of specific ERP features of interest reveals no significant differences between the ERP amplitudes of the waveforms generated using the two techniques, and Pearson correlation suggests a high degree of similarity at the individual level (0.88 for N100, 0.78 for P300, and 0.80 for N400, all p<; 0.05). Conclusion: The new denoising technique is capable of operating on individual-channel EEG data, and produces results that are similar to those produced by conventional denoising techniques that use data from large whole-head electrode arrays. This new approach may thus enable more widespread use of ERP techniques in real world settings with low-density electrode arrays. |
|---|
| Publication date | 2020-09-12 |
|---|
| Publisher | IEEE |
|---|
| In | |
|---|
| Language | English |
|---|
| Peer reviewed | Yes |
|---|
| Export citation | Export as RIS |
|---|
| Report a correction | Report a correction (opens in a new tab) |
|---|
| Record identifier | 7eeb7f2b-4c9e-40f1-a8da-13fb2baad63c |
|---|
| Record created | 2022-02-14 |
|---|
| Record modified | 2022-02-14 |
|---|
