Cross Site Projects

Treatment Outcomes. We examined the effects of three different types of treatment (sentence processing, naming, and spelling), administered at each of our three sites (Boston University, Johns Hopkins and Northwestern University) on a group of 100 participants with chronic-stroke induced aphasia. Individuals with aphasia were randomly assigned to receive either a 12-week long course of linguistically-based treatment or to serve as control participants. Results indicate significant improvement for the treatment group only, at all the three sites. All three treatment groups also performed significantly better on follow-up (3 months post-treatment) compared to baseline, but poorer compared to immediately post-treatment (See Figure 1). Overall, these findings indicate that all three types of treatment resulted in significantly improved performance on the main dependent measure used at each site, and that treatment outcomes were mostly maintained three months after completion of treatment.

Right Hemisphere (RH) Grey Matter (GM) Density was examined in aphasic (n=40) and cognitively healthy (age-matched) (n=40) participants to better understand right hemisphere (RH) recruitment for language recovery in people with aphasia. We (Lukic et al., 2017) used voxel-based morphometry (VBM) to evaluate GM density across RH regions in both participant groups. Results showed reduced GM volume in right Supplementary Motor Area (SMA), but increased volume in right MTG in the aphasic participants compared to the healthy controls. The amount of GM in both regions was positively correlated with language production and comprehension scores, respectively. These findings suggest that the RH may support functions previously performed by left hemisphere regions and have important implications for understanding post-stroke reorganization.

Perfusion. Through this experiment we (Thompson et al., 2017) examined perfusion values in 35 aphasic and 16 healthy control participants. Region-of-interest (ROI) analyses showed a varied pattern of perfusion in regions across both hemispheres in the aphasic participants. Eight ROIs in the left hemisphere (LH) and seven in the right hemisphere (RH) showed reduced perfusion values in aphasic individuals (compared to healthy participants). Furthermore, one region in the LH showed greater perfusion in aphasic participants than in healthy individuals. However, there were no significant correlations between perfusion values and language abilities in these regions. Perfusion values were significantly below normal in the left 0-6mm ring surrounding the lesion, in line with previous research that shows reduced perfusion in regions that are closest to the lesion (Brumm et al, 2010; Fridriksson et al., 2012; Richardson et al., 2011). (Figure 2).

Reliability of BOLD signal activation (IntraClass Correlation (ICC) analyses). For this study, we used an fMRI story comprehension task where participants were presented with blocks of auditory stories (control condition = reversed speech) or blocks of written stories (control condition = false fonts) (Figure 3). Participants were required to listen to or read the stories and were asked questions off-line about the content. We then computed ICC values across scans, comparing baseline scans 1 and 2 and post-testing (three-months) scans 1 and 2 in both patients (n=65) and a cohort of healthy controls (n=22). Healthy participants’ ICC coefficients indicated excellent reliability, while individuals with aphasia showed moderate reliability at both time points.

Resting State Analyses (rsMRI). In this experiment, still in progress, we asked healthy and aphasic participants at all three sites to lie still in the scanner while looking at a white cross on a black background. Voxel-wise fractional amplitude of low frequency fluctuations (fALFF) values were extracted from all the 96 regions-of-interest (ROIs) included in the Harvard-Oxford atlas (Desikan et al., 2006). Preliminary results showed a significant interaction of Group by ROI, indicating differences between the patients and healthy controls at multiple ROIs. fALFF values at baseline were also found to be predictive of responsiveness to treatment at each site. Response to sentence processing (Northwestern) and naming (Boston) treatments was predicted by baseline fALFF values in the Supplementay Motor Area (SMA) in the right hemisphere (RH). Spelling treatment (Johns Hopkins) outcomes were predicted by baseline fALFF values in the Inferior Frontal Gyrus and in the Orbitofrontal Cortex in the left hemisphere (LH). Finally, resting state connectivity within right hemisphere (RH) temporal areas was predictive of response to both spelling and naming treatments. Overall, these preliminary findings suggest that resting state connectivity within both LH and RH may predict response to treatment in individuals with chronic aphasia. 

See our recent publications!

Figure 1. Pre- to post-treatment performance for the Tx and NH groups, and follow-up for the Tx group on domain-specific language tasks. BU=Boston (naming), JH=Johns Hopkins (spelling), NU=Northwestern (sentence comprehension and production).
Figure 2. Perilesional perfusion values in the LH and RH (Thompson et al., 2017).
Figure 3: Story comprehension task activation maps (healthy N=22 and stroke N=65, p=0.05 whole-brain corrected). Color bars show z-score ranges. Slices show x=51/-51.