Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Dual-Stream Language Network Mapping in People with Post-Stroke Aphasia and Age-Matched Controls Using Functional Near Infrared Spectroscopy
Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Erin Meier1, Veronica Fletcher1, Caela Hung1, Pratham Khare1, Priyansh Khare1, Esprit Ange Andraos1, Leanna Ugent1, Gengchen Wei1; 1Northeastern University
Introduction: There are comparatively few functional neuroimaging studies focused on early aphasia recovery, likely due in part to the logistical difficulties of using fMRI with this population. While functional Near-Infrared Spectroscopy (fNIRS) is more cost-effective than fMRI and can be administered across diverse settings, its utility for language mapping in people with aphasia (PWA) remains underexplored. To assess the potential use of fNIRS to index neurological recovery from aphasia, we aimed 1) to determine whether fNIRS activation patterns from semantic and phonological tasks map onto ventral and dorsal perisylvian brain regions, respectively[1,2], and 2) to evaluate how brain activity in PWA varies by lesion location and task performance. Method: Thirteen PWA (2 women, 11 men; mean age: 65.8 +/- 6.7 years) and 15 neurologically healthy adults (6 women, 9 men; mean age: 62.5 +/- 11.4 years) completed one run of the Adaptive Language Mapping Semantic Matching[3] and Rhyme Judgment[4] tasks. Both tasks followed an AB design with alternating 20s blocks of experimental (semantic or rhyme judgment) and control (letter string matching) trials. fNIRS data were acquired with two connected 8x8 NIRx NIRSport2s, resulting in 44 long and 8 short channels symmetrically positioned over bilateral perisylvian language areas. fNIRS data preprocessing was completed in Homer3[5] using standard steps. For each task, changes in oxygenated hemoglobin (HbO) for experimental>control were modeled within a GLM (t=-2 to 25s). Results: For Semantic Matching, controls significantly activated left inferior frontal gyrus pars opercularis (LIFGop) and triangularis (LIFGtri) and left mid to posterior superior (LSTG), middle (LMTG), and inferior temporal (LITG) gyri. As a group, PWA significantly activated mid LMTG only for Semantic Matching. For Rhyme Judgment, controls significantly activated LIFGtri, left supramarginal gyrus (LSMG), and left angular gyrus (LAG) whereas PWA significantly recruited LSMG only. A significant group x task effect on experimental trial accuracy was observed (F(1, 52)=4.46, p=0.040) with PWA exhibiting significantly lower accuracy on Rhyme Judgment experimental trials than controls (t=4.52, p<0.001) with no between-group differences in Semantic Matching accuracy (t=1.53, p=0.132). No significant relationships between task accuracy and activation were found using LASSO regression. Across PWA, greater damage to LIFGtri (r=0.68) and LSTG (r=0.66) was associated with increased activity within LMFG for Semantic Matching. For Rhyme Judgment, greater damage to LIFGtri and LIFGop was associated with widespread reductions in activation, whereas greater damage to LSMG coincided with increased activation across most right hemisphere regions. Discussion: In controls, activation was primarily observed in ventral stream regions for Semantic Matching and dorsal stream regions for Rhyme Judgment, as predicted according to the dual-route model of language processing[1,2]. As expected, substantial intersubject variability in activation was seen across PWA, yet peak areas of activity for each task still aligned with the dual-route model. The lack of significant brain-behavior relationships were likely in part due to the adaptative nature of the tasks, which reduced the variance in task accuracy. Our results suggest that when critical nodes network nodes are damaged, PWA resort to recruiting ipsilesional regions typically associated with domain-general processing[6] or right hemisphere areas.
Topic Areas: Disorders: Acquired,