Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

EEG Evidence of Low-Level Speech Processing in Severe Brain Injury

Poster E33 in Poster Session E, Saturday, October 8, 3:15 - 5:00 pm EDT, Millennium Hall

Parul Jain¹, Mary Conte², Henning Voss^3,4, Jonathan Victor^2,5, Nicholas Schiff^2,5; ¹Weill Cornell Graduate School of Medical Sciences, New York, NY, ²Feil Family Brain and Mind Research Institute, Weill Cornell Medical College, New York, NY, ³College of Human Ecology, Cornell University, Ithaca, NY, ⁴Department of Radiology, Weill Cornell Medicine, New York, NY, ⁵Department of Neurology, New York Presbyterian Hospital, New York, NY

Assessing cognitive function – especially language processing – in severely brain-injured patients is critical for prognostication, care, and development of communication devices (e.g. brain-computer interfaces). In patients with compromised motor function, EEG measures have been used to probe language processing at the level of comprehension and at the very lowest level – tracking of the natural speech envelope (NSE) – but not at the intermediate level of phoneme processing. Here, we develop and apply methods to identify a differential phoneme-class specific response (DPR) and NSE tracking, and apply them in parallel to severely brain-injured patients. We studied a cohort of 26 severely brain-injured patients (17 traumatic, 5 anoxic, 4 other etiologies; 19 M) and 10 healthy controls. Patients’ level of function was assessed via the Coma Recovery Scale – Revised (CRS-R) at bedside, and EEG/fMRI measured of command following via tests of motor imagery. For the present paradigm, EEG was recorded using an augmented 10-20 montage (37 electrodes), during presentation of a 148 s audio clip of Alice in Wonderland. EEG responses to phoneme classes (approximants, fricatives, nasals, plosives, and vowels) were extracted by averaging the EEG (2-15 Hz band-pass filtered) with respect to time markers placed at phoneme onset. Tracking of NSE was assessed from the same recordings, and was measured by cross-correlating 2 s segments of EEG response with the speech envelope. For the phonemic analysis, healthy controls had DPRs for multiple phoneme pairs, which agrees with previous phoneme processing studies done in healthy controls (Khalighinejad et. al., 2017). The response was observed during the entire 500 ms analysis interval, most often at 200-300 ms after phoneme onset. Most differences localized to left frontal, central parieto-occipital, right temporal, and right frontal regions. DPRs were found in all patients, but patients with evidence of command following had an early bilateral response in the first 50 ms. In comparison, patients without any evidence of command following lacked the early response but had a late response around 250 ms. For NSE tracking, responses in the first 100 ms distinguished patients without any evidence of language processing. Specifically, patients without any evidence of language processing had a frontopolar response in the first 100 ms and a late fronto-central response around 200 ms. Patients with evidence of language processing had a more global response in the first 100 ms. Finally, we noted that the overall dynamics of the DPR and NSE tracking in healthy controls was largely similar but there was a critical difference in the spatial pattern: the central parieto-occipital response observed in DPRs was absent in NSE tracking response. In conclusion, we developed an EEG-based method to probe low-level speech processing in severely brain-injured patients. When applied to assess NSE tracking and phoneme processing, our results suggest that patients with evidence of command following have an early bilateral response component, which is absent or weak for patients without any evidence of command following.

Topic Areas: Perception: Speech Perception and Audiovisual Integration, Methods