Poster Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions

Mapping Language-Related Brain Regions in Early Development via Integration of fMRI and fNIRS

Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House

Sara Sanchez-Alonso¹, Rebecca Canale², Isabel Nichoson³, Virginia Chambers¹, Richard Aslin¹; ¹Yale University, ²University of Connecticut, ³Tulane University

Introduction: The neural mechanisms supporting language learning in early development remain poorly understood, partly due to the challenges of imaging the developing brain with functional magnetic resonance imaging (fMRI)—the gold-standard technique for studying adult neurobiology. To bridge this gap, we have developed a multi-modal imaging protocol that integrates fMRI and functional near-infrared spectroscopy (fNIRS). fNIRS is a more versatile tool to study brain function in young children due to its portability, ease of implementation in naturalistic settings and robustness to motion relative to fMRI. Our approach, combined with a child-friendly localizer, enables mapping of language-related brain regions in early development. Methods: Study 1 (data collection complete): We collected simultaneous fMRI (Siemens 3T PRISMA) and fNIRS signals in 40 young adults (18- to 35-year-olds) using a previously validated language localizer task. Study 2 (ongoing): fMRI-fNIRS signals (either simultaneous or concurrent) are currently being collected from toddlers and pre-school-age children (n=10, target n=25) using a naturalistic language localizer task. This naturalistic paradigm consists of 20-sec child-friendly video clips that contrast speech and nonspeech conditions. Data preprocessing: The fNIRS data are preprocessed using NeuroDOT (Eggebrecht & Culver, 2019) to create anatomically registered maps of cerebral hemodynamics. Image reconstruction steps include mapping fNIRS optode locations to the individual anatomical MR images, implementing the light model, and running 3D image reconstruction on CIFTI surface maps. In addition, we used eight short-separation fNIRS channels to measure and remove extracerebral signals. The fMRI data are preprocessed via Human Connectome Project (HCP) pipelines (Glasser et al., 2013). For scalp-to-anatomy co-registration of the fNIRS signal, we use a direct mapping of optode locations to the T1w image via vitamin E capsules on the fNIRS caps at each optode location. A key innovation is co-registration of fNIRS channels to each individual’s cortical surface (i.e., a CIFTI surface map), and co-registration of fMRI voxels using the HCP MMP 1.0 atlas (Glasser et al., 2016) and subsequent cortical parcellation, which consists of 180 cortical areas per hemisphere. Results: Results from the adult sample (Study 1) show two main findings: 1) similar to fMRI, fNIRS signals recruit canonical language regions in fronto-temporal cortices with lateralization to the left hemisphere; and 2) Optode-to-anatomy co-registration highlights the importance of precise optode localization (e.g., head shape, head circumference). Ongoing quantitative analyses of the two signals consist of assessing the timeseries temporal correlation across fMRI and fNIRS as a function of cortical region, location of the fNIRS probes, and implementation of different smoothing parameters during fMRI data preprocessing. Preliminary data from Study 2 shows bilateral engagement of language areas and modulation by the level of language knowledge. Conclusion: Our study provides a comparison of fMRI and fNIRS signals collected simultaneously from the same subjects and shows that fNIRS can be used to investigate cortical activations underlying speech processing during a traditional language localizer task. Preliminary data using a child-friendly, naturalistic paradigm supports the feasibility of fNIRS for mapping language areas in early development.

Topic Areas: Language Development/Acquisition, Methods