PP1308 - Direct Brain Recordings of Speech in Noise in Children with Intractable Epilepsy

Abstract: Most investigations of how childrens’ brains process speech in noise are limited to behavioral or noninvasive measures such as EEG or ABR. To better understand the cortical regions contributing to speech in noise processing in children, we recorded intracranial signals from children being treated for drug-resistant epilepsy while they performed the BKB-SIN speech in noise task. Behavioral and neural responses were analyzed based on SNR level and correct/incorrect responses to view how auditory cortex is activated during the task. We found stronger neural responses in the superior temporal gyrus (STG) to higher SNR sentences and for correct versus incorrect trials.

Summary: Direct recordings of brain activity during speech in noise processing are rare because they may only be obtained in specific populations, such as patients with epilepsy undergoing surgical monitoring. Scalp EEG and ABR can show how brain regions are engaged over time but have little spatial specificity. Intracranial recordings, on the other hand, allow direct recording of neural activity at high spatiotemporal resolution. While such recordings have been used to study speech in noise processing in adults (Khalighinejad et al., 2019; Patel et al. 2022), few studies have investigated this in children. Defining how the auditory cortex represents speech in noisy environments in children has important implications for understanding development of typical hearing as well as auditory processing disorders. Previous behavioral research found that +15 SNR is best for cortical and behavioral processing of speech in noise (Benítez-Barrera et al. 2021) in children. Additionally, younger children have poorer performance on speech in noise tasks than young adults (Buss et al. 2019, Sobon et al. 2019). Consonant identification in noise may not reach adult-like performance until 14 years of age (Johnson, 2000). Here, we take advantage of rare intracranial recordings in patients undergoing surgical treatment of epilepsy to investigate speech in noise processing. We administered the Bamford-Kowal-Bench Speech in Noise (BKB-SIN) test to children and young adults while neural responses were recorded from intracranial electrodes placed around the auditory cortex. We predicted that there would be greater variability in the amplitude of neural responses for differing SNR levels in Heschl’s gyrus (primary auditory cortex) compared to lateral STG. We also expected larger response amplitudes for correct compared to incorrect trials in the lateral STG. We recorded data from 7 participants (3 females, 4 males; ages 7-20 years old, mean = 11.7 years; median = 8). Patients were recruited from Dell Children’sMedical Center in Austin, Texas, and Texas Children’s Hospital in Houston, Texas, and had intracranial electrodes placed in their brains as part of surgical treatment of intractable epilepsy. We recorded neural signals while participants listened to the BKB-SIN lists using free field via speakers at the end of the patient’s hospital bed, calibrated to ~70 dB SPL. Behavioral and neural responses were analyzed based on SNR level and correct/incorrect responses. Neural data was aligned to sentence timings and preprocessed to remove epileptiform activity. After performing a common average reference, we extracted the high gamma power (70-150 Hz), which correlates with multi-unit activity. We then computed event related potentials locked to sentence onset. Sentences with high SNR (21, 18, 15 dB) were compared to those with mid SNRs (12, 9, 6 dB) and low SNRs (3, 0, -3, -6 dB). BKB-SIN scores ranged from -1 to 17, with a mean of 4.55. We found that the high SNRs had stronger neural onset responses than mid and low, and that both SNR and correct/incorrect performance were encoded in STG activity. To our knowledge, this is the first study to directly record neural activity in the auditory cortex of children during SIN tasks.