EEG Selective Attention Decoding for CI Users
The cochlear implant (CI) acts as a kind of artificial cochlea, transforming the acoustic signals, captured by a microphone into electric pulses, thereby bypassing the damaged structures of the ear and directly stimulating the auditory nerve. Most CI users obtain good speech understanding in the absence of background noise. However, CI users still face difficulties in understanding speech in more challenging listening environments with multiple speakers, background noise and reverberation (i.e., the cocktail party problem). In such situations, normal-hearing (NH) listeners can focus on one target speaker and effectively suppress other present speakers. This effect is called selective attention. The ability to selectively attend is probably impaired in CI users and may be one of the reasons for the limitations in speech understanding in challenging listening environments.
This project investigates the possibilities of decoding selective attention in NH listeners and CI users. It has been shown that neural activity in the cerebral cortex, especially in the delta (1–4 Hz) and theta (4–8 Hz) frequency bands, tracks the amplitude envelope of a complex auditory stimulus such as speech. Therefore, electroencephalography (EEG) data can be used to track an attended speech source in NH listeners and CI users using high-density EEG caps. The selective attention can be further decoded by designing spatio-temporal filters, also called decoder, which serve to reconstruct the speech from the EEG data. The goal of this project is to investigate different types of decoders, including linear forward/backward models or non-linear models such as deep neural networks. Moreover, different methods to record and further to decode selective attention are investigated. Promising recording systems are the cEEGrid system, which represents portable EEG recording device or invasive electrode located on the temporal lobe.
This project is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy - EXC 2177/1 - Project number 390895286.
|Head of Research Group:
||[Prof. Dr.-Ing. Waldo Nogueira]|
||DHZ-Deutsches HörZentrum Hannover
|Phone:||+49 (0)511 532 8025|
|Fax:||+49 (0)511 532 6833|
Nogueira, W., Dolhopiatenko, H., Schierholz, I., Büchner, A., Mirkovic, B., Bleichner, M. G., & Debener, S. (2019). Decoding selective attention in normal hearing listeners and bilateral cochlear implant users with concealed ear EEG. Frontiers in Neuroscience. Frontiers in Neuroscience, 13(JUL), 1–15. https://doi.org/10.3389/fnins.2019.00720
Nogueira, W., Cosatti, G., Schierholz, I., Egger, M., Mirkovic, B., and Büchner, A. (2020). Toward Decoding Selective Attention From Single-Trial EEG Data in Cochlear Implant Users. IEEE Transactions on Biomedical Engineering, vol. 67, no. 1, pp. 38-49, Jan. 2020, doi: 10.1109/TBME.2019.2907638.
Nogueira, W. and Dolhopiatenko, H. (2020). Towards Decoding Selective Attention from Single-Trial EEG Data in Cochlear Implant users based on Deep Neural Networks," ICASSP 2020 - 2020 IEEE International Conference on Acoustics. Speech and Signal Processing (ICASSP), Barcelona, Spain, 2020, pp. 8708-8712, doi: 10.1109/ICASSP40776.2020.9054021.