Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Large-Scale Functional Connectivity During Language Processing in Post-Stroke Aphasia
Poster Session B, Wednesday, September 30, 4:30 - 6:30 pm, Wangari Maathai
Essiet-Adidiong Ette1, Maria Varkanitsa1, Swathi Kiran1; 1Boston University
Post-stroke aphasia (PSA) is a communication disorder caused by damage to regions of the left hemisphere involved in language. Though the structural damage is localized, the functional damage may span beyond the lesion site and disrupt widespread networks. Functional network connectivity (FNC) analysis enables characterization of how activity between brain regions differs across tasks and conditions. Examining these connectivity patterns can help clarify how language processing is preserved, reorganized, or disrupted after stroke. Forty individuals with chronic PSA participated in the experiment, which included structural and functional MRI protocols. During task-based fMRI, all subjects completed various language-dependent and cognitive tasks presented in a block design, including reading (sentences vs. nonwords), listening (intact vs. degraded), arithmetic (easy vs. hard), movie watching, and story listening tasks, with conditions varying in demand or difficulty. Lesions were masked out using ITK-SNAP. Further data preparation and analyses were done using CONN toolbox. Data were preprocessed with a pipeline that included smoothing, functional realignment & unwarping, slice‑timing correction, outlier identification, and segmentation/normalization to MNI space. Regions of interest (ROIs) included Schaefer atlas parcels, subject-specific lesion masks, and standard CONN-derived grey matter, white matter, and CSF masks. Group-level ROI-to-ROI analyses examining task- and condition-specific connectivity patterns across large-scale networks, including the frontoparietal, language, default mode, visual, and salience networks, among others, were conducted. These analyses showed significant functional connections across multiple networks within each task and condition (connection: p < .01; cluster: p < .05). However, direct condition contrasts for the tasks identified significant differences for the LangListen task only, comparing intact and degraded speech. The degraded condition was associated with positive connectivity between salience and frontoparietal networks, as well as within-frontoparietal networks, both suggesting that participants relied more heavily on executive and attentional systems when the speech signal was harder to understand. Intact speech showed stronger negative connectivity among visual network components, indicating reduced cross-coupling or segregation of visual regions when the auditory stimulus requires little additional top-down support. Although the remaining tasks did not show significant condition differences at the cluster level, their condition-specific maps were not identical. There was qualitative variation in network-pair significance and connectivity direction, indicating that some instances of reorganization are more subtle than others. These results imply that language processing in post-stroke aphasia depends on interactions across several large-scale brain networks. During this process, not only is the language network engaged, but non-linguistic networks, such as the salience, frontoparietal, and visual networks, are also dynamically recruited. Changes in connectivity within and between these networks appear to track the task's perceptual demand and are critical for understanding how language functions in the brain. Further analysis, examining individual differences within each ROI can provide a detailed account of the connectivity changes. At the individual level, different neural markers can be identified that may be specific to aphasia severity. Contextualizing both individual and group-level analyses can help inform and improve individualized treatments and recovery trajectories for individuals with PSA. Ultimately, this research can inform hypotheses about adaptive versus maladaptive network reorganization after stroke.
Topic Areas: Disorders: Acquired, Computational Approaches