GK
G. Khosrokhavar
info
Please Note
<p>This page displays the records of the person named above and is not linked to a unique person identifier. This record may need to be merged to a profile.</p>
1 records found
1
The Acoustic Change Complex (ACC) is a cortical auditory Event-Related Potential (ERP) used as an objective measure of auditory discrimination. Its clinical adoption is limited by the long recording times required to recover the ACC waveform from noisy single-trial EEG through conventional trial averaging.
This thesis develops a two-part single-trial analysis pipeline. Part 1 classifies each trial as Clear ACC, Possible ACC, or Noise Dominant using rule-based hard-gate filtering followed by template-based scoring. Part 2 corrects per-trial latency offsets using four alignment algorithms: direct N1-peak alignment, direct P2-peak alignment, the Woody filter, and dynamic time warping (DTW).
Evaluated on seven normal-hearing adults at electrode E6, the pipeline raises the averaged N1–P2 amplitude from 2.8 µV (grand average of all trials) to 9.9 µV for the best alignment method, a 3.5× improvement. The Woody filter yields the highest absolute amplitude, while direct P2-peak alignment yields the highest noise-corrected amplitude. Validation against non-hearing acquisitions yields a mean specificity of 96.4%, confirming that the detection pipeline does not produce substantial spurious detections on stimulus-absent EEG.
This work contributes toward a more efficient ACC recording framework by showing that trial-level information can be used to enhance waveform recovery and reduce dependence on long averaging protocols. ...
This thesis develops a two-part single-trial analysis pipeline. Part 1 classifies each trial as Clear ACC, Possible ACC, or Noise Dominant using rule-based hard-gate filtering followed by template-based scoring. Part 2 corrects per-trial latency offsets using four alignment algorithms: direct N1-peak alignment, direct P2-peak alignment, the Woody filter, and dynamic time warping (DTW).
Evaluated on seven normal-hearing adults at electrode E6, the pipeline raises the averaged N1–P2 amplitude from 2.8 µV (grand average of all trials) to 9.9 µV for the best alignment method, a 3.5× improvement. The Woody filter yields the highest absolute amplitude, while direct P2-peak alignment yields the highest noise-corrected amplitude. Validation against non-hearing acquisitions yields a mean specificity of 96.4%, confirming that the detection pipeline does not produce substantial spurious detections on stimulus-absent EEG.
This work contributes toward a more efficient ACC recording framework by showing that trial-level information can be used to enhance waveform recovery and reduce dependence on long averaging protocols. ...
The Acoustic Change Complex (ACC) is a cortical auditory Event-Related Potential (ERP) used as an objective measure of auditory discrimination. Its clinical adoption is limited by the long recording times required to recover the ACC waveform from noisy single-trial EEG through conventional trial averaging.
This thesis develops a two-part single-trial analysis pipeline. Part 1 classifies each trial as Clear ACC, Possible ACC, or Noise Dominant using rule-based hard-gate filtering followed by template-based scoring. Part 2 corrects per-trial latency offsets using four alignment algorithms: direct N1-peak alignment, direct P2-peak alignment, the Woody filter, and dynamic time warping (DTW).
Evaluated on seven normal-hearing adults at electrode E6, the pipeline raises the averaged N1–P2 amplitude from 2.8 µV (grand average of all trials) to 9.9 µV for the best alignment method, a 3.5× improvement. The Woody filter yields the highest absolute amplitude, while direct P2-peak alignment yields the highest noise-corrected amplitude. Validation against non-hearing acquisitions yields a mean specificity of 96.4%, confirming that the detection pipeline does not produce substantial spurious detections on stimulus-absent EEG.
This work contributes toward a more efficient ACC recording framework by showing that trial-level information can be used to enhance waveform recovery and reduce dependence on long averaging protocols.
This thesis develops a two-part single-trial analysis pipeline. Part 1 classifies each trial as Clear ACC, Possible ACC, or Noise Dominant using rule-based hard-gate filtering followed by template-based scoring. Part 2 corrects per-trial latency offsets using four alignment algorithms: direct N1-peak alignment, direct P2-peak alignment, the Woody filter, and dynamic time warping (DTW).
Evaluated on seven normal-hearing adults at electrode E6, the pipeline raises the averaged N1–P2 amplitude from 2.8 µV (grand average of all trials) to 9.9 µV for the best alignment method, a 3.5× improvement. The Woody filter yields the highest absolute amplitude, while direct P2-peak alignment yields the highest noise-corrected amplitude. Validation against non-hearing acquisitions yields a mean specificity of 96.4%, confirming that the detection pipeline does not produce substantial spurious detections on stimulus-absent EEG.
This work contributes toward a more efficient ACC recording framework by showing that trial-level information can be used to enhance waveform recovery and reduce dependence on long averaging protocols.