You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

 
Poster D32, Thursday, November 9, 6:15 – 7:30 pm, Harborview and Loch Raven Ballrooms

Using background connectivity to index recovery of function in acquired language impairments

Yuan Tao1, Brenda Rapp1;1Johns Hopkins University

It is both theoretically and clinically important to understand the neural changes that take place in the course of recovery of function in the adult lesioned brain. From a theoretical perspective, this is part of the larger enterprise of understanding the possibilities and constraints of neuroplasticity. From a clinical perspective, this understanding contributes to our ability to optimize neural interventions for recovery of function following brain lesions. Recovery of function, as a form of (re)learning, can be expected to result in local neural changes as well as changes in the interactions between more distant brain regions (Buchel & Friston, 1999). In this research, we examine changes in the neural interactions that accompany recovery of function subsequent to rehabilitation for stroke-induced written language deficits. Background connectivity, as developed by Norman-Haignere et al. (2011), provides a measure of the interaction between brain areas in a way that removes the contribution of stimulus-locked activation fluctuations. We used classification-based analysis approaches to detect rehabilitation-related changes in background connectivity patterns. Eleven right-handed (5 females) individuals with acquired dysgraphia subsequent to a single left-hemisphere stroke (14-103 months post-stroke) underwent biweekly behavioral rehabilitation for approximately 12 weeks. All individuals exhibited significant improvement in spelling accuracy and fMRI data were collected during performance of a spelling task both before and after treatment. After standard pre-processing, we estimated background connectivity using GLM residuals. The GLMs included task events, 6 motion regressors, and averaged sub-cortical time-series to remove physiological artefacts. For 264 5mm-radius spheres (locations from Power et al., 2011), we calculated the absolute values of all pairwise Pearson correlations (Fisher z-transformed). Lesioned voxels were assigned a zero value, indicating no information. We applied Linear Discriminate Analysis (LDA) to classify each individual’s pre vs. post-training correlation matrices using a leave-one-subject-out cross-validation procedure. Prior to each cross-validation iteration, feature selection was carried out to identify the most informative connections. Pre- vs. post-training correlation matrices were most reliably distinguished with 200 connections (accuracy=0.73, permutation test p=0.01) and 51 of these were identified in all cross-validation iterations. Thirty-seven exhibited a correlation decrease from pre- to post-training and only 12 showed an increase (two-tailed p<0.05). Among the connections with decreased correlations, more than 87% were either within the left hemisphere (40%) or between hemispheres, while only 13% were within the right hemisphere. Although informative connections were widely distributed, nodes involved in connectivity decreases were concentrated in bilateral motor cortex and the anterior cingulate/paracingulate cortex, while connectivity increases were concentrated in the right anterior insular, contralateral to the most heavily lesioned area. In conclusion, the connectivity-based approach revealed systematic neural changes from pre to post treatment for acquired language deficits in both the directionality of the interaction (increase or decrease connectivity) and the hemispheric lateralization. The finding of widespread functional connectivity decreases is consistent with previous findings of abnormally high levels of post-stroke resting state connectivity (Carter et al., 2012; Park et al., 2011) in certain brain areas. Furthermore, the results suggest that rehabilitation may support normalization of left hemisphere connectivity.

Topic Area: Language Disorders

Back to Poster Schedule