Brain network topology influences response to intensive comprehensive aphasia treatment

Abstract

BACKGROUND:

Recent imaging studies indicate that aphasia is associated with large-scale reorganization of brain networks. Today, neuroimaging studies show that various brain connectivity properties, as measured by resting state fMRI, can partially explain different behavioral symptoms in and across various patient groups. Despite these observations, the neural networks underlying the progress and recovery of aphasia following intensive treatment remains relatively obscure.

OBJECTIVE:

To examine the role of brain network properties in determining recovery of aphasia following intensive therapy in stroke patients.

METHODS:

We studied eight patients with left hemispheric lesions who completed an intensive comprehensive aphasia program (ICAP). Language and cognition were assessed before and after four weeks of intensive treatment. In addition, all patients underwent resting state fMRI prior to and after treatment. We used graph theory analysis to evaluate relationships of baseline brain network properties, such as efficiency, modularity, and connectivity to clinical improvements.

RESULTS:

We found global properties such as efficiency and interhemispheric connectivity could partially explain recovery. More importantly, we identified two unique brain networks that are significantly associated with improvement in language and attention related behavior.

CONCLUSIONS:

These results suggest baseline brain functional properties play a key role in determining responsiveness of patients with aphasia to intensive comprehensive aphasia treatment. Furthermore, these results indicate that brain mechanisms underlying language comprehension and processes are different from those involved in spatial attention.

Keywords

Aphasia stroke resting-state fMRI brain networks intensive treatment

1 Introduction

Aphasia is a language disorder resulting from brain injury that affects more than one million people in the United states alone (Ellis, Dismuke, & Edwards, 2010). Research regarding neuroplasticity in animal models has indicated that intensity of treatment following induced injury can result in significant changes in excitability, mapping reorganization, and dendritic and axonal sprouting (Nudo, 2011). However, Nudo (2011) has emphasized that the amount of treatment dosing in animal studies is vastly different from dosing in human rehabilitation treatment. For example, animal studies have included up to 600 repetitions of a task in one day as compared to 32 repetitions in a functional upper extremity task for humans. Although evidence overall for intensive aphasia treatment has been equivocal over two decades of research (Brady, Kelly, Godwin, Enderby, & Campbell, 2016; Cherney, Patterson, Raymer, Frymark, & Schooling, 2010; Cherney, Patterson, Raymer, Frymark, & Schooling, 2008; Cherney, Patterson, & Raymer, 2011), several studies have demonstrated greater gains with more intensive treatment (Bakheit, Shaw, Carrington, & Griffiths, 2007; Barthel, Meinzer, Djundja, & Rockstroh, 2008; Bhogal, Teasell, Speechley, & Albert, 2003; Bhogal, Teasell, Foley, & Speechley, 2003; Breitenstein et al., 2017; Kurland, Baldwin, & Tauer, 2010; Meinzer et al., 2004; Poeck, Huber, & Willmes, 1989). Furthermore, a Cochrane Review of randomized control trials of speech-language therapy for aphasia found that functional communication was significantly better for those who received high intensity, high dose, or a long duration of therapy compared with lower intensity, lower dose, or a shorter period of time (Brady et al., 2016).

One mechanism for delivering intensive therapy is through an Intensive Comprehensive Aphasia Program (ICAP). ICAPs have increased in response to the evidence supporting intensive treatment and the needs of patients seeking treatment for their aphasia (Rose, Cherney, & Worrall, 2013). Rose et al. (2013) defined an ICAP as a program that provides intensive, individualized aphasia treatment up to 6 hours a day for up to 5–6 weeks to a cohort of participants who start and end the program at the same time. The comprehensive component of an ICAP is addressed by providing different modes of treatment (individual, computer-based, and group treatment), focusing on both the impairment and activity-participation aspects of the aphasia, and offering caregiver education (Chapey et al., 2000; Eames, Hoffmann, Worrall, & Read, 2010; Elman & Bernstein-Ellis, 1999; Wallace et al., 2016). Results from clinical programs and research studies have shown that this treatment delivery platform is effective for the majority of participants with improvements demonstrated on behavioural language measures and patient-reported outcome measures (Babbitt, Worrall, & Cherney, 2015, 2016; Breitenstein et al., 2017; Code, Torney, Gildea-Howardine, & Willmes, 2010; Dignam et al., 2015; Hoover, Caplan, Waters, & Carney, 2016; Persad, Wozniak, & Kostopoulos, 2013; Rodriguez et al., 2013; Winans-Mitrik et al., 2014). There is also limited but inconclusive research that identifies factors which may predict recovery in aphasia (Babbitt et al., 2016; El Hachioui et al., 2013; Lambon Ralph, Snell, Fillingham, Conroy, & Sage, 2010; Pedersen, Vinter, & Olsen, 2003; Persad et al., 2013; Plowman, Hentz, & Ellis, 2012; Watila & Balarabe, 2015). However, the neural mechanisms underlying such improvements remain largely unknown.

Over the last two decades, functional magnetic resonance imaging (fMRI) has emerged as an invaluable non-invasive tool for studying and understanding brain functional activity and organization in health and disease. While task-specific fMRI allows us to detect changes in brain function related to a specific task, resting-state fMRI (rs-fMRI) measures the coherence (correlation) in intrinsic activity between brain regions in the absence of an overt task, and therefore allows us to examine the natural functional organization of the brain and its subsequent changes in various neurological conditions including stroke, Alzheimer’s disease, various neuropsychiatric disorders and epilepsy (for review see Fox and Greicius (2010)). In contrast to task-specific fMRI, rs-fMRI scans can be easily obtained from patients, even if they suffer from severe aphasia, and result in robust connectivity properties that are reliable across scans and institutions (Biswal et al., 2010). Over the last decade, rs-fMRI has been used to examine changes in functional connectivity of language networks and their modulation in response to therapy in a small number of studies (Klingbeil, Wawrzyniak, Stockert, & Saur, 2017). While changes in interhemispheric connectivity appear to be consistently related to language deficits, the functional reorganization underlying aphasia remains relatively largely unknown. This is mainly due to the differences in the analytical approaches used and heterogeneity of patient populations, as well as the lack of a generalized theoretical framework (Klingbeil et al., 2017).

Table 1
Demographic information for eight participants

Subject Age Months Post-Onset Gender Type of Aphasia WAB-R –AQ score

Pat 01 75 16 Male Fluent 31.7

Pat 02 58 17 Female Non-Fluent 23.6

Pat 03 48 20 Female Non-Fluent 48.0

Pat 04 62 4 Male Fluent (Anomic) 85.7

Pat 05 78 12 Male Fluent (Anomic) 86.4

Pat 06 48 11 Female Non-Fluent 43.9

Pat 07 67 32 Male Non-Fluent 33.4

Pat 08 30 4 Male Fluent (Anomic) 84.1

AVG 55.2 13.9 54.6

STDV 17.4 8.7 26.6

Subject	Age	Months Post-Onset	Gender	Type of Aphasia	WAB-R –AQ score
Pat 01	75	16	Male	Fluent	31.7
Pat 02	58	17	Female	Non-Fluent	23.6
Pat 03	48	20	Female	Non-Fluent	48.0
Pat 04	62	4	Male	Fluent (Anomic)	85.7
Pat 05	78	12	Male	Fluent (Anomic)	86.4
Pat 06	48	11	Female	Non-Fluent	43.9
Pat 07	67	32	Male	Non-Fluent	33.4
Pat 08	30	4	Male	Fluent (Anomic)	84.1
AVG	55.2	13.9			54.6
STDV	17.4	8.7			26.6

The classical framework for studying and treating aphasia is based on the view that specific behavioral and clinical symptoms in patients are caused by localized brain regions with distinct functional specialization (Dronkers, Wilkins, Van Valin, Redfern, & Jaeger, 2004). For example, speech output deficits are thought to be primarily caused by damage to the insular and frontal cortex, while reading deficits are related to injury within temporal regions (Dronkers et al., 2004). This symptom-lesion view is mainly based on studies in small number of patients that consider each brain region independently (Bates et al., 2003; Dronkers et al., 2004), and therefore mainly ignore large-scale brain organization and its role in behavioral symptoms onset, progression and modulation in response to therapy. Today there is increased evidence showing that human brains exhibit topological attributes in common with other complex systems that provide the functional basis for sensory and cognitive information processing (Bullmore & Sporns, 2009; Fornito, Zalesky, & Bullmore, 2010), and that are essential for acquiring experiences and learning (Lewis, Baldassarre, Committeri, Romani, & Corbetta, 2009). Furthermore, converging evidence from various clinical studies show that disruption of large-scale network properties may play a key role in the emergence and manifestation of behavioural symptoms in multiple neurological and psychiatric disorders (Bullmore & Sporns, 2009; Catani & ffytche, 2005).

Within this framework, we used rs-fMRI to characterize various local and global brain network properties in eight participants with aphasia and investigated their role in determining clinical outcomes following an intensive comprehensive aphasia program. We hypothesized that brain network topology plays an important contributory role in determining the capacity for improvement of aphasia symptoms following therapy. We used advanced and well validated graph theoretical approaches to construct brain networks regional cortical and subcortical nodes, with edges (or links) drawn between nodes to represent their functional correlation. Brain networks were constructed and studied at different thresholds resulting in sparse but fully connected networks. To insure that outcomes were meaningful, all network properties were compared in relation to a large offsite control data set (n = 750).

2 Methods

2.1 Participants

Eight first-time participants in an ICAP elected to take part in this study. Written informed consent and medical clearance for the MRI scans were obtained. Of the eight subjects, five were male. Ages ranged from 30 to 78 years (mean = 55.2; SD = 17.4) and average months post-onset ranged from 4 to 32 months (mean = 13.9; SD = 8.7). Typical language and cognitive evaluation tasks were administered during pre- and post-treatment evaluation sessions. The subjects’ initial scores on the Western Aphasia Battery – Revised Aphasia Quotient (WAB-R AQ) (Kertesz, 2007) ranged from 31.7 to 86.4 (mean = 55.9; SD = 25.1). Four subjects were characterized by non-fluent aphasia and four with fluent aphasia. Of those with fluent aphasia, three presented with an anomic aphasia (see Table 1). The study was approved by the Northwestern University Institutional Review Board. In addition, 736 off-site healthy control subjects (328 females, 408 males; average age: mean = 31.94, range = 18 – 71 years, s.d. = 3.14 years) were procured from Connectome1000 (https://www.nitrc.org/projects/fcon_1000/).

2.2 Study design and treatment

2.2.1 Behavioural measures

A full battery of tests were administered as part of the ICAP. All eight participants completed pre- and post-treatment behavioural language and attention measures. These measures included the complete WAB-R, Boston Naming Test (BNT) (Kaplan, Goodglass, & Weintraub, 2001), and the Conners’ Continuous Performance Test II (CPT II) (Conners & Staff, 2000). The CPT II is a nonverbal measure of attention (Riccio, Reynolds, Lowe, & Moore, 2002). A change in the confidence interval score was used in this analysis to evaluate improvement in overall attentional processing.

2.2.2 Treatment

During their participation in the clinical ICAP, subjects received treatment for six hours a day, five days a week for four weeks for a total of 120 hours. The therapy program was based on evidence-based treatments that were individualized based on the participants’ interests, goals, and specific language and communication deficits. Each day comprised two hours of one-on-one speech and language therapy with an experienced speech and language pathologist, and an hour each of intensive language action therapy, computer-based treatment, a supported conversation group, and a reading/writing class. Treatments included but were not limited to: semantic feature analysis (Boyle & Coelho, 1995), treatment of underlying forms (Thompson, Shapiro, Kiran, & Sobecks, 2003), constraint-induced language therapy (Difrancesco, Pulvermüller, & Mohr, 2012), verb network strengthening treatment (Edmonds, Nadeau, & Kiran, 2009), computerized script training (Cherney, Halper, Holland, & Cole, 2008), computerized Oral Reading for Language in Aphasia (ORLA) (Cherney, 2010), writing treatments (Beeson, 1999) and conversation groups (Elman & Bernstein-Ellis, 1999). Family and caregiver education at the beginning and end of the program was also included. For more detailed information about the ICAP, see Babbittet al. (2015).

2.2.3 Scanning parameters for patients

For all patients, MPRAGE type T1-anatomical brain images were acquired with a 3T Siemens Trio whole-body scanner with echo-planar imaging (EPI) capability using the standard radio-frequency head coil with the following parameters: voxel size 1×1×1 mm; TR = 2500 ms; TE = 3.36 ms; flip angle = 9°; in-plane matrix resolution = 256×256; slices = 176; field of view = 256 mm. Rs-fMRI images were acquired on the same day with the following parameters: Multi-slice T2*-weighted echo-planar images with repetition time TR = 2.2 s, echo time TE = 30 ms, flip angle = 90°, number of slices = 38, slice thickness = 3 mm, in-plane resolution = 64×64, number of volumes was 240. The 38 slices covered the whole brain from the cerebellum to the vertex. Two rs-fMRI images and one T1-anatomical image were collected per patient.

2.2.4 Anatomical data preprocessing and analysis

Lesion masks were identified and drawn on the high-resolution T1-weighted structural scans by two personnel independently. Only voxels identified by the two investigators were included in the final subject mask. Brains were then registered to standard MNI space using linear registration in FSL (FLIRT). Lesion localizations were identified using the Harvard-Oxford structural atlases (HOSA; http://www.cma.mgh.harvard.edu/fsl_atlas.html). Lesion size for each region of interest were computed as the percent of lesion voxels relative to the total number of voxels of the region.

2.2.5 Resting-state fMRI preprocessing

The pre-processing of each subject’s time series of fMRI volumes was performed using the FMRIB Expert Analysis Tool (FEAT (Smith et al., 2004), www.fmrib.ox.ac.uk/fsl) and encompassed: Discarding the first five volumes to allow for magnetic field stabilization; skull extraction using BET; slice time correction; motion correction. One major concern in the study is that head motion artefacts might contribute to differences in connectivity. Motion and several sources of noise, which may contribute to non-neuronal fluctuations, were removed from the data through a robust strategy based on volume censoring, which has been shown to reduce group differences due to motion to chance levels (Power et al., 2014). Finally, the cleaned time series were band-pass filtered (0.01–0.1 Hz) to identify the low frequency fluctuations of interest.

2.2.6 Brain graph construction

Nodes of the brain functional network were defined using a parcellation schema containing 264 regions that are 10-mm diameter spheres centered on the coordinates reported previously (Power et al., 2011) that were identified using both a multiple task fMRI meta-analyses (Dosenbach et al., 2010) and a cross-validated rs-fMRI functional connectivity parcellation technique (Cohen et al., 2008). To construct the whole brain networks for each subject, we first computed the Pearson correlation coefficient (R) for all possible pairs of ROIs time series from the preprocessed rs-fMRI data. Since, we acquired 2 scans per subject at each time point, we averaged the 2 resultant matrices to produce one 264 by 264 correlation matrix for each subject. For each subject, the threshold was calculated to produce a fixed number of edges (M) to be able to compare the extracted graphs (Achard et al., 2012). Therefore, the values of the threshold are subject-dependent. Each of these extracted graphs comprised of N = 264 nodes and M undirected edges corresponding to the significant nonzero absolute values of correlation greater than the value of the threshold. Since the value of the chosen threshold is important (Sophie Achard & Bullmore, 2007; Achard et al., 2012), we chose to test several values of threshold, from a conservative threshold corresponding to 2% connection density (the percentage of edges with respect to the maximum number of possible edges [(N×N-1)/2]) to a lenient threshold corresponding to 10% link density. Networks constructed at 10% link density are dubbed sparse networks, while those constructed at 50% are dense.

2.2.7 Graph metrics calculations

Following the methods described by Bullmore and Sporns (2012), several topological properties of the graphs were computed using the Brain Connectivity Toolbox (BCT, https://sites.google.com/site/bctnet/). These measurements include global efficiency (a measure of information integration) and modularity (a global measure of the near-decomposability of the network into a community structure of sparsely interconnected modules). The measurement of network modularity can be computed using various algorithms and different approaches (Alexander-Bloch et al., 2012). Here we used a similar approach to that described by Cole, Bassett, Power, Braver, and Petersen (2014). First, we used a fast and accurate multi–iterative generalization of the Louvain algorithm (Blondel, Guillaume, Lambiotte, & Lefebvre, 2008) that is provided and recommended by the BCT. We used this technique to compute a single unitary measure of modularity (between 0 and 1) that described the hierarchical organization of the network. Highly structured networks with well-defined partitions will yield values closer to 1, while random networks will yield values closer to 0. Since correlation networks can exhibit modularity degeneracy (the existence of multiple distinct community structures of the same network), we computed modularity over 100 repetitions. The final modularity for any given network was then determined as the average modularity of the 100 repetitions. Network partitions (communities) were not identified from data presented in this study. Instead,community memberships of the 264 ROIs were assigned based on previous results (Power et al., 2011) that were validated across two independent data sets and various thresholds (Power et al., 2011). Interhemispheric connectivity was simply computed as the total number of edges between all left and right ROIs. Inter-modular connectivity was determined as the percent of number of connections of all ROIs between two given modules relative to the total number of possible connections.

2.2.8 Off-site healthy controls data processing and analysis

Resting-state fMRI data pre-processing, graph construction and graph metric calculation for the 736 off-site healthy controls were performed using identical methods to those described for the patients in the above sections.

3 Results

3.1 Behavior changes in response to therapy

A battery of clinically relevant and standardized tests were used to assess language and attention abilities. Significant changes in behavioral scores between the two visits were assessed using a paired t-test. Overall, patients showed significant improvements across most language and attentional domains (Table 2). Boston naming test (BNT), Aphasia Quotient (AQ), WAB-R repetition (Rep), Raven’s Progressive Matrixes (RAV) and Connor’s Continuous Performance Test (CPT-2) showed significant changes following treatment (all p < 0.05). WAB-R Reading (Rdg) and Writing (Wtg) scores showed trends (p = 0.116 and p = 0.073 respectively).

Table 2
Patients show improved outcome across multiple behavioral domains

Behavioral Language Testing Visit 1 Visit 2 t-score p-value

BNT / 60 20.87±23.32 26.75±24.75 −2.82 0.025

WAB-R – AQ / 100 54.60±26.57 62.60±27.46 −7.84 0.0001

WAB-R – Rdg / 100 59.50±23.77 68.25±26.20 −1.79 0.116

WAB-R – Wtg / 100 49.75±31.35 57.06±28.16 −2.11 0.073

WAB-R – Rep / 100 54.26±3.53 64.75±26.26 −3.35 0.012

RCPM / 37 30.37±2.72 32.13±2.03 −3.86 0.006

CPT II CI 71.90±22.83 56.32±20.11 +4.22 0.003

Behavioral Language Testing	Visit 1	Visit 2	t-score	p-value
BNT / 60	20.87±23.32	26.75±24.75	−2.82	0.025
WAB-R – AQ / 100	54.60±26.57	62.60±27.46	−7.84	0.0001
WAB-R – Rdg / 100	59.50±23.77	68.25±26.20	−1.79	0.116
WAB-R – Wtg / 100	49.75±31.35	57.06±28.16	−2.11	0.073
WAB-R – Rep / 100	54.26±3.53	64.75±26.26	−3.35	0.012
RCPM / 37	30.37±2.72	32.13±2.03	−3.86	0.006
CPT II CI	71.90±22.83	56.32±20.11	+4.22	0.003

Data presented as mean±s.d. Statistical significances were determined by two-tailed paired t-test. Boston Naming Test –BNT; Western Aphasia Battery-Revised –WAB-R; Aphasia Quotient –AQ; WAB-R Reading Subtest –Rdg; WAB-R Writing Subtest –Wtg; WAB-R Repetition Subtest –Rep; Ravens Colored Progressive Matrices –RCPM; Conners’ Continuous Performance Test II Confidence Interval –CPT II CI (100% = disordered, 50% = normal).

We used principal component analysis (PCA), a common data reduction strategy that identifies hidden factors that capture the possible association of behavioral scores across subjects. We found that all behavioral measurement at baseline (visit 1) clustered into two behavioral factors that explained 84.87% of the total variance of the data. The first factor, labelled AQ, comprised all language related processes including the BNT, AQ, Rep, Rdg and Wtg. The second factor CPT included only the CPT II score (Fig. 1a). Similar to visit 1 scores, behavioral data following treatment (visit 2) loaded on the two factors and explained 87.94% of the data variance (Fig. 1b). These results suggest that a common central mechanism may underlie multiple language related performances, and that is probably independent from the mechanism involved in attention related tasks. Given these results, we only investigated the brain mechanisms underlying AQ (representative of Factor 1) and CPT (representative of Factor 2) (Fig. 1c). Note that similar results would have been obtained using any other language variables, since all language behavioral scores showed a correlation >0.9.

Fig.1

Correlation matrices and factor analyses of behavioural scores. (a) Heat map (left) show the correlation across all behavioral scores at baseline (visit 1). Table (right) shows principle component analysis. Two main factors were identified, factor 1 contained all language domains, AQ, while factor 2 contained CPT. RAV loaded equally on both factors. (b) Behavior scores after therapy (visit 2) shows similar associations to baseline. (c) Scatter plot shows the correlation between behavior changes in AQ (factor 1) and CPT (factor 2). Despite both behavioral measures showing signifecant improvement, there were no correlation between them. Higher scores indicate better improvement.

3.2 Lesion size and location show minimum effect in determining treatment outcome

We investigated the extent to which anatomical and functional brain properties at Visit 1 predicted treatment outcome (Visit 2 – Visit 1) using asimple correlation model. All lesions were localized to the left hemisphere (Fig. 2a) and exhibited various sizes ranging from 101.48 to 215.76 cm³. Lesion size showed no significant relation to AQ (r = 0.23, p = 0.64) or CPT outcome (r = −0.11, p = −0.84). In addition to total lesion size, we investigated the relationship of regional damage (lesion size) to outcome using a region of interest analysis. Percentage lesion size for various cortical and subcortical regions were computed and submitted to a correlation analysis. Overall, we observed no significant correlations between regional brain damage and AQ and CPT changes following treatment. It is worthy to note that despite the absence of significant results (probably due to small number of patients), we observed that the CPT and AQ changes showed some correlation to different brain regions (Fig. 2b). Overall, global and local anatomical brain damage exhibited poor association with improvements of both language and attentional symptoms in aphasia patients.

Fig.2

Anatomical localization demonstrates no effect on treatment outcomes. (a) Lesion overlay map in standard MNI space. Color scale represents the number of subjects with lesions at each voxel. All lesions were constricted to the left hemisphere. (b) Bar graph (left) show mean±s.d. of lesion size for 48 cortical and 6 subcortical regions. Scatter plot (right) shows the correlation of lesion size with Factor 1 AQ (black) and Factor 2 CPT (gray) improvements. There were no significant associations between lesion localization and treatment outcome. (Red lines represent significance, p < 0.05 uncorrected).

3.3 Brain global functional connectivity properties play an important contributory role in determining treatment outcome

We investigated the role of brain functional properties in determining outcome responses. Brain graphs were generated using functional connectivity estimated between each pair of the 264 functional brain regions of interest (Fig. 3 a–d). Several key global network properties including global efficiency, modularity and interhemispheric connectivity were computed and examined in relationship to the off-site healthy group and treatment outcomes. Global efficiency is a measure of information integration and is defined as the average inverse shortest path length between all pairs of nodes in a given network. Therefore, networks with higher efficiency are able to transfer information between spatially distributed regions at lower costs (i.e. shorter paths). Modularity measures the extent of division of a network into unique non-overlapping modules or communities. Different modules can provide a source of functional specialization (e.g. vision, attention, etc.), while their interconnectivity is required for integration of information across different functional domains. Both specialization and integration are needed in order to sustain proper functionality of the human brain. Interhemispheric connectivity measures the extent of functional connectivity between the right and left hemispheres of the brain.

Patients showed global efficiencies (similar to those of a healthy population (t-value = 0.45, p = 0.95, one-tailed t-test) (Fig. 4a). Relationship between efficiency and outcome was assessed using a Pearson correlation analysis. Efficiency was significantly correlated to AQ (p < 0.05), but not CPT (p = 0.23) improvements (Fig. 4b). Modularity in patients was also similar to that of the healthy subjects (t-value = 1.24, p = 0.63, one-tailed t-test) (Fig. 4c) and did not correlate with AQ (p = 0.13) or CPT (p = 0.23) changes following treatment (Fig. 4d). Finally, interhemispheric connectivity was significantly decreased in patients compared to the healthy distribution (Fig. 4e), and was only associated with CPT improvements (p < 0.05) (Fig. 4f).

Fig.3

Graph theoretical brain network partition and properties. (a) Network partition of 260 functional nodes (ROI) described previously. The eleven major networks (modules) are labelled on the right (Power et al., 2011). (b) Heat maps represent the group mean RSFC correlation matrices for the 260 nodes in healthy (left) and patients (right). Nodes belonging to the same module showed higher correlations. (c) Top view of group mean RSFC brain network for healthy subjects (network density = 5% ). (d) Individual RSFC brain networks for all patients (network density = 5% ).

Fig.4

Global brain network properties are significantly related to treatment outcomes. (a) Left plot show mean±s.d. of global efficiency in patients (green) and healthy subjects (gray) for 2% , 4% , 6% , 8% and 10% network density analyses. Line plot shows the individual global efficiencies, normalized to healthy mean and s.d., for all densities. A unitary normalized score for each patient was generated by averaging normalized scores across all densities (right circles). Overall, patients exhibited similar global efficacy compared to healthy (mean = −0.35, red bar). (b) Scatter plots depict the relationship between global efficiency and treatment outcome. Global efficiency showed a significant association with improved AQ, but not CPT II. Similar analyses were performed for modularity (c and d) and IHC (e and f). Modularity showed no changes in patients and was not correlated with treatment outcomes. IHC was significantly lower in patient compared to healthy subjects (mean±s.d. = −1.72±0.34; t-score = −4.98, p < 0.01, two-tailed unpaired t-test) and was related to CPT II improvement.

3.4 Language and attention treatment outcomes are driven by specific network connectivity

In addition to global properties, we examined whether functional connectivity between distinct brain networks (modules or communities) play a role in determining outcome responses. Inter-modular connectivity for 5% density brain networks were generated by 1) computing the total number of connections between any two given modules and normalizing it by the total number of possible connections (Fig. 5a). Association between modular connectivity and behavioural changes was assessed using a simple correlation analysis. We observed that AQ improvement showed a very significant dependence on connectivity between the default-mode network (DMN) and auditory regions (Fig. 5b). CPT improvement showed a significant positive correlation with connectivity strength of the salience network (SAN) and visual areas (Fig. 5c).

Fig.5

Treatment outcomes are driven by specific network connectivity. (a) Modularity based connectivity maps were constructed by adding all ROI connections between any given pair of Modules and normalizing by the maximum number of connections. (b) Scatter plots show the modular connectivity (density = 5% ) that was significantly (p < 0.05, FDR corrected) related to treatment. Improvement in AQ scores was uniquely correlated to connectivity between DMN and auditory network, while Improvement in CPT II was related to increased connectivity between SAN and visual network. Inserts show the connectivity across all densities.

Taken together, these results show that improvement of aphasia symptoms following an intensive comprehensive aphasia program can be explained in part by baseline brain connectivity properties. Furthermore, different improvements across behavioural domains are dependent on both global and system-specific connectivity.

4 Discussion

The aims of this paper were to investigate the influence of baseline anatomical and functional brain properties in response to intensive treatment delivered during an ICAP. Using resting state functional connectivity (rsFC) is an important shift from task-based fMRI research as it is well-suited to examine changes in whole brain activity versus specific, task-based representative imaging. Resting state functional connectivity specifically allows for examining subjects with a broad range of deficits and studying multiple networks simultaneously (Carter, Shulman, & Corbetta, 2012). Here we showed that baseline rsFC properties in persons with aphasia were strongly associated with treatment outcomes whereas other factors such as lesion location and size were not associated with outcomes. Large-scale properties like global efficiency and interhemispheric connectivity were related to improved language and visual attention performances respectively. We also were able to identify specific connectivity changes within discrete networks that may play an integral role in functional improvement. Connectivity between the DMN and auditory regions were highly related to improved language, while connectivity between salience network and visual regions was related to increase visual attention performance. To our knowledge, this constitutes the first rs-fMRI study that identifies specific rsFC properties that predict outcome in persons with aphasia following treatment.

Some early studies have found that post-treatment for aphasia there are network changes. Marcotte, Perlbarg, Marrelec, Benali, and Ansaldo (2013) found that there was more integration in the default mode network (DMN) which resembled networks of normal controls. They also noted that better pre-treatment integration may predict outcomes following semantic-based treatment.

One research study examining resting state functional connectivity (rsFC) of language networks in healthy individuals found more extensive language networks than task-based fMRIs were able to identify (Tomasi & Volkow, 2012). Systems neuroscience applications in aphasia show that rsFC measures offer information about multiple networks, are ideal for studying remote physiological effects of lesions on distant areas, and provide a direct measure of temporal coherence of regional interaction (Carter et al., 2012; Kiran, 2012; Turken, U, & Dronkers, 2011; Ulm, Copland, & Meinzer, 2016). A recent stroke recovery and rehabilitation roundtable identified rsFC as one biomarker which was named a priority for developing prediction of outcome and treatment response in aphasia research (Boyd et al., 2017). Klingbeil et al. (2017) describe the benefits and limitations of using rsFC to examine language networks. One significant benefit is that rsFC scans can be completed with persons with a range of aphasia severity, which is important because ICAPs generally accept a heterogeneous group of participants with a wide-range of severity levels, different types of aphasia, and varying times post-onset.

Exploration of resting state network changes post-treatment is limited to a small number of studies in the aphasia literature. Klingbeil et al. (2017) have reviewed the literature and identified four studies which examined rsFC following treatment. In response to naming treatment, van Hees et al. (2014) found that after treatment participants demonstrated more normalized connectivity in the language networks in the left hemisphere. Duncan and Small (2016) found that narrative discourse improvements following treatment corresponded to increased modularity of global networks. Although these studies found changes in rsFC which correlated with improvements due to treatment, there isinsufficient evidence to conclusively state whether those changes are a return to normal patterns or represent compensatory patterns (Marangolo et al., 2016).

4.1 Limitations of current study

This study included only eight subjects who self-selected to take part in an ICAP and were willing to undergo rs-fMRI scans. The subjects demonstrated a wide range of type and severity of aphasia and time post-onset. The small sample size and the heterogeneity across subjects make it difficult to generalize findings. In addition, there were no mechanisms in place to ensure fidelity of treatment across subjects since the ICAP was a clinical program in which patients each received highly individualized treatment based on their specific deficits. However, this is a starting point to begin to understand recovery from aphasia following intensive therapy.

4.2 Concluding statement and future directions

Future research will include larger numbers of subjects to identify whether subgroups can be determined based on variables such as type and severity of aphasia or lesion location. Time post-onset and age also have been identified as factors which may contribute to improvements following treatment in an ICAP (Babbitt et al., 2016) and these factors may influence brain network typology and the changes that occur following treatment. In light of recent work indicating potential superiority of distributed treatment on measures of impairment (Dignam et al., 2015), it will be important to compare rs-fMRI in participants during intensive vs distributed treatment programs.

Conflict of interest

None to report.

Footnotes

Acknowledgments

We thank Dave Copland at the University of Queensland for early discussions regarding resting state analysis for intensive aphasia treatment. We also thank Todd Parrish and Xue Wang at Northwestern University for providing scanning time at the Center for Advanced MRI.

References

Achard, S., & Bullmore, E. (2007). Efficiency and Cost of Economical Brain Functional Networks. PLoS Computational Biology, 3(2), e17. doi:10.1371/journal.pcbi.0030017

Achard, S., Delon-Martin, C., Vertes, P.E., Renard, F., Schenck, M., Schneider, F.,… Bullmore, E.T. (2012). Hubs of brain functional networks are radically reorganized in comatose patients. Proceedings of the National Academy of Sciences, 109(50), 20608-20613. doi: 10.1073/pnas.1208933109

Alexander-Bloch, A., Lambiotte, R., Roberts, B., Giedd, J., Gogtay, N., & Bullmore, E. (2012). The discovery of population differences in network community structure: New methods and applications to brain functional networks in schizophrenia. NeuroImage, 59(4), 3889-3900. doi:10.1016/j.neuroimage.2011.11.035

Babbitt, E.M., Worrall, L., & Cherney, L.R. (2015). Structure, Processes, and Retrospective Outcomes From an Intensive Comprehensive Aphasia Program. American Journal of Speech-Language Pathology, 24(4), S854. doi:10.1044/2015_ajslp-14-0164

Babbitt, E.M., Worrall, L., & Cherney, L.R. (2016). Who Benefits From an Intensive Comprehensive Aphasia Program?Topics in Language Disorders, 36(2), 168-184. doi:10.1097/tld.0000000000000089

Bakheit, A.M.O., Shaw, S., Carrington, S., & Griffiths, S. (2007). The rate and extent of improvement with therapy from the different types of aphasia in the first year after stroke. Clinical Rehabilitation, 21(10), 941-949. doi:10.1177/0269215507078452

Barthel, G., Meinzer, M., Djundja, D., & Rockstroh, B. (2008). Intensive language therapy in chronic aphasia: Which aspects contribute most?Aphasiology, 22(4), 408-421. doi:10.1080/02687030701415880

Bates, E., Wilson, S.M., Saygin, A.P., Dick, F., Sereno, M.I., Knight, R.T., & Dronkers, N.F. (2003). Voxel-based lesion–symptom mapping. Nature Neuroscience, 6(5), 448-450. doi:10.1038/nn1050

Beeson, P.M. (1999). Treating acquired writing impairment: Strengthening graphemic representations. Aphasiology, 13(9-11), 767-785. doi:10.1080/026870399401867

10.

Bhogal, S.K., Teasell, R., Speechley, M., & Albert, M.L. (2003). Intensity of Aphasia Therapy, Impact on Recovery * Aphasia Therapy Works!Stroke, 34(4), 987-993. doi:10.1161/01.str.0000062343.64383.d0

11.

Bhogal, S.K., Teasell, R.W., Foley, N.C., & Speechley, M.R. (2003). Rehabilitation of Aphasia: More Is Better. Topics in Stroke Rehabilitation, 10(2), 66-76. doi:10.1310/rcm8-5tul-nc5d-bx58

12.

Biswal, B.B., Mennes, M., Zuo, X.-N., Gohel, S., Kelly, C., Smith, S.M.,… Colcombe, S. (2010). Toward discovery science of human brain function. Proceedings of the National Academy of Sciences, 107(10), 4734–4739.

13.

Blondel, V.D., Guillaume, J.-L., Lambiotte, R., & Lefebvre, E. (2008). Fast unfolding of communities in large networks. Journal of Statistical Mechanics: Theory and Experiment, 2008(10), P10008. doi:10.1088/1742-5468/2008/10/p10008

14.

Boyd, L.A., Hayward, K.S., Ward, N.S., Stinear, C.M., Rosso, C., Fisher, R.J.,… Cramer, S.C. (2017). Biomarkers of stroke recovery: Consensus-based core recommendations from the Stroke Recovery and Rehabilitation Roundtable. International Journal of Stroke, 12(5), 480–493. doi:10.1177/1747493017714176

15.

Boyle, M., & Coelho, C.A. (1995). Application of Semantic Feature Analysis as a Treatment for Aphasic Dysnomia. American Journal of Speech-Language Pathology, 4(4), 94. doi:10.1044/1058-0360.0404.94

16.

Brady, M.C., Kelly, H., Godwin, J., Enderby, P., & Campbell, P. (2016). Speech and language therapy for aphasia following stroke. Cochrane Database of Systematic Reviews(6). doi:10.1002/14651858.CD000425.pub4

17.

Breitenstein, C., Grewe, T., Flöel, A., Ziegler, W., Springer, L., Martus, P.,… Haeusler, K.G. (2017). Intensive speech and language therapy in patients with chronic aphasia after stroke: A randomised, open-label, blinded-endpoint, controlled trial in a health-care setting. The Lancet, 389(10078), 1528–1538.

18.

Bullmore, E., & Sporns, O. (2009). Complex brain networks: Graph theoretical analysis of structural and functional systems. Nature Reviews Neuroscience, 10(3), 186–198. doi:10.1038/nrn2575

19.

Bullmore, E., & Sporns, O. (2012). The economy of brain network organization. Nature Reviews Neuroscience, 13(5), 336–349. doi:10.1038/nrn3214

20.

Carter, A.R., Shulman, G.L., & Corbetta, M. (2012). Why use a connectivity-based approach to study stroke and recovery of function?NeuroImage, 62(4), 2271–2280. doi:10.1016/j.neuroimage.2012.02.070

21.

Catani, M., & ffytche, D.H. (2005). The rises and falls of disconnection syndromes. Brain, 128(10), 2224–2239. doi:10.1093/brain/awh622

22.

Chapey, R., Duchan, J.F., Elman, R.J., Garcia, L.J., Kagan, A., Lyon, J.G., & Simmons Mackie, N. (2000). Life Participation Approach to Aphasia: A Statement of Values for the Future. The ASHA Leader, 5(3), 4–6. doi:10.1044/leader.FTR.05032000.4

23.

Cherney, L., Patterson, J., Raymer, A., Frymark, T., & Schooling, T. (2010). Updated evidence-based systematic review: Effects of intensity of treatment and constraint-induced language therapy for individuals with stroke-induced aphasia. Available at asha.org.

24.

Cherney, L.R. (2010). Oral Reading for Language in Aphasia (ORLA): Evaluating the Efficacy of Computer-Delivered Therapy in Chronic Nonfluent Aphasia. Topics in Stroke Rehabilitation, 17(6), 423–431. doi:10.1310/tsr1706-423

25.

Cherney, L.R., Halper, A.S., Holland, A.L., & Cole, R. (2008). Computerized Script Training for Aphasia: Preliminary Results. American Journal of Speech-Language Pathology, 17(1), 19. doi:10.1044/1058-0360(2008/003)

26.

Cherney, L.R., Patterson, J.P., Raymer, A., Frymark, T., & Schooling, T. (2008). Evidence-Based Systematic Review: Effects of Intensity of Treatment and Constraint-Induced Language Therapy for Individuals With Stroke-Induced Aphasia. Journal of Speech Language and Hearing Research, 51(5), 1282. doi:10.1044/1092-4388(2008/07-0206)

27.

Cherney, L.R., Patterson, J.P., & Raymer, A.M. (2011). Intensity of Aphasia Therapy: Evidence and Efficacy. Current Neurology and Neuroscience Reports, 11(6), 560–569. doi:10.1007/s11910-011-0227-6

28.

Code, C., Torney, A., Gildea-Howardine, E., & Willmes, K. (2010). Outcome of a One-Month Therapy Intensive for Chronic Aphasia: Variable Individual Responses. Seminars in Speech and Language, 31(01), 021–033. doi:10.1055/s-0029-1244950

29.

Cohen, A.L., Fair, D.A., Dosenbach, N.U.F., Miezin, F.M., Dierker, D., Van Essen, D.C.,… Petersen, S.E. (2008). Defining functional areas in individual human brains using resting functional connectivity MRI. NeuroImage, 41(1), 45–57. doi:10.1016/j.neuroimage.2008.01.066

30.

Cole, Michael W., Bassett

Danielle S.

, Power

Jonathan D.

, Braver

Todd S.

, & Petersen

Steven E.

(2014). Intrinsic and Task-Evoked Network Architectures of the Human Brain. Neuron, 83(1), 238–251. doi:10.1016/j.neuron.2014.05

31.

Conners, C.K., & Staff, M. (2000). Conners’ continuous performance Test II (CPT II v. 5).

32.

Difrancesco, S., Pulvermüller, F., & Mohr, B. (2012). Intensive language-action therapy (ILAT): The methods. Aphasiology, 26(11), 1317–1351. doi:10.1080/02687038.2012.705815

33.

Dignam, J., Copland, D., McKinnon, E., Burfein, P., O'Brien, K., Farrell, A., & Rodriguez, A.D. (2015). Intensive Versus Distributed Aphasia Therapy. Stroke, 46(8), 2206–2211. doi:10.1161/strokeaha.115.009522

34.

Dosenbach, N.U.F., Nardos, B., Cohen, A.L., Fair, D.A., Power, J.D., Church, J.A.,… Schlaggar, B.L. (2010). Prediction of Individual Brain Maturity Using fMRI. Science, 329(5997, 1358–1361. doi:10.1126/science.1194144

35.

Dronkers, N.F., Wilkins, D.P., Van Valin, R.D., Redfern, B.B., & Jaeger, J.J. (2004). Lesion analysis of the brain areas involved in language comprehension. Cognition, 92(1–2), 145–177. doi:10.1016/j.cognition.2003.11.002

36.

Duncan, E.S., & Small, S.L. (2016). Increased Modularity of Resting State Networks Supports Improved Narrative Production in Aphasia Recovery. Brain Connectivity, 6(7), 524–529. doi:10.1089/brain.2016.0437

37.

Eames, S., Hoffmann, T., Worrall, L., & Read, S. (2010). Stroke Patients’ and Carers’ Perception of Barriers to Accessing Stroke Information. Topics in Stroke Rehabilitation, 17(2), 69–78. doi:10.1310/tsr1702-69

38.

Edmonds, L.A., Nadeau, S.E., & Kiran, S. (2009). Effect of Verb Network Strengthening Treatment (VNeST) on lexical retrieval of content words in sentences in persons with aphasia. Aphasiology, 23(3), 402–424. doi:10.1080/02687030802291339

39.

El Hachioui, H., Lingsma, H.F., van de Sandt-Koenderman, M.W., Dippel, D.W., Koudstaal, P.J., & Visch-Brink, E.G. (2013). Long-term prognosis of aphasia after stroke. J Neurol Neurosurg Psychiatry, 84(3), 310–315. doi:10.1136/jnnp-2012-302596

40.

Ellis, C., Dismuke, C., & Edwards, K.K. (2010). Longitudinal trends in aphasia in the United States. NeuroRehabilitation, 27(4), 327–333.

41.

Elman, R., & Bernstein-Ellis, E. (1999). Psychosocial Aspects of Group Communication Treatment - Preliminary Findings. Seminars in Speech and Language, 20(01), 65–72. doi:10.1055/s-2008-1064009

42.

Elman, R.J., & Bernstein-Ellis, E. (1999). The Efficacy of Group Communication Treatment in Adults With Chronic Aphasia. Journal of Speech Language and Hearing Research, 42(2), 411. doi:10.1044/jslhr.4202.411

43.

Fornito, A., Zalesky, A., & Bullmore, E.T. (2010). Network scaling effects in graph analytic studies of human resting-state FMRI data. Frontiers in Systems Neuroscience, 4, 22.

44.

Fox, M.D., & Greicius, M. (2010). Clinical Applications of Resting State Functional Connectivity. Frontiers in Systems Neuroscience, 4, 19. doi:10.3389/fnsys.2010.00019

45.

Hoover, E.L., Caplan, D.N., Waters, G.S., & Carney, A. (2016). Communication and quality of life outcomes from an interprofessional intensive, comprehensive, aphasia program (ICAP). Topics in Stroke Rehabilitation, 24(2), 82–90. doi:10.1080/10749357.2016.1207147

46.

Kaplan, E., Goodglass, H., & Weintraub, S. (2001). Boston naming test: Pro-ed.

47.

Kertesz, A. (2007). WAB-R : Western Aphasia Battery-Revised. San Antonio, TX: PsychCorp.

48.

Kiran, S. (2012). What Is the Nature of Poststroke Language Recovery and Reorganization?ISRN Neurology, 2012, 1–13. doi:10.5402/2012/786872

49.

Klingbeil, J., Wawrzyniak, M., Stockert, A., & Saur, D. (2017). Resting-state functional connectivity: An emerging method for the study of language networks in post-stroke aphasia. Brain and Cognition. doi:10.1016/j.bandc.2017.08.005

50.

Kurland, J., Baldwin, K., & Tauer, C. (2010). Treatment-induced neuroplasticity following intensive naming therapy in a case of chronic Wernicke's aphasia. Aphasiology, 24(6-8), 737–751. doi:10.1080/02687030903524711

51.

Lambon Ralph, M.A., Snell, C., Fillingham, J.K., Conroy, P., & Sage, K. (2010). Predicting the outcome of anomia therapy for people with aphasia post CVA: Both language and cognitive status are key predictors. Neuropsychological Rehabilitation, 20(2), 289–305. doi:10.1080/09602010903237875

52.

Lewis, C.M., Baldassarre, A., Committeri, G., Romani, G.L., & Corbetta, M. (2009). Learning sculpts the spontaneous activity of the resting human brain. Proceedings of the National Academy of Sciences, 106(41), 17558–17563. doi:10.1073/pnas.0902455106

53.

Marangolo, P., Fiori, V., Sabatini, U., De Pasquale, G., Razzano, C., Caltagirone, C., & Gili, T. (2016). Bilateral Transcranial Direct Current Stimulation Language Treatment Enhances Functional Connectivity in the Left Hemisphere: Preliminary Data from Aphasia. Journal of Cognitive Neuroscience, 28(5), 724–738. doi:10.1162/jocn_a_00927

54.

Marcotte, K., Perlbarg, V., Marrelec, G., Benali, H., & Ansaldo, A.I. (2013). Default-mode network functional connectivity in aphasia: Therapy-induced neuroplasticity. Brain and Language, 124(1), 45–55. doi:10.1016/j.bandl.2012.11.004

55.

Meinzer, M., Elbert, T., Wienbruch, C., Djundja, D., Barthel, G., & Rockstroh, B. (2004). Intensive language training enhances brain plasticity in chronic aphasia. BMC Biology, 2(1), 20.

56.

Nudo, R.J. (2011). Neural bases of recovery after brain injury. Journal of Communication Disorders, 44(5), 515–520. doi:10.1016/j.jcomdis.2011.04.004

57.

Pedersen, P.M., Vinter, K., & Olsen, T.S. (2003). Aphasia after Stroke: Type, Severity and Prognosis. Cerebrovascular Diseases, 17(1), 35–43. doi:10.1159/000073896

58.

Persad, C., Wozniak, L., & Kostopoulos, E. (2013). Retrospective Analysis of Outcomes from Two Intensive Comprehensive Aphasia Programs. Topics in Stroke Rehabilitation, 20(5), 388–397. doi:10.1310/tsr2005-388

59.

Plowman, E., Hentz, B., & Ellis, C. (2012). Post-stroke aphasia prognosis: A review of patient-related and stroke-related factors. Journal of Evaluation in Clinical Practice, 18(3), 689–694.

60.

Poeck, K., Huber, W., & Willmes, K. (1989). Outcome of Intensive Language Treatment in Aphasia. Journal of Speech and Hearing Disorders, 54(3), 471. doi:10.1044/jshd.5403.471

61.

Power, Jonathan

, Cohen, Alexander

, Nelson, Steven

, Wig, Gagan

, Barnes, Kelly

, Church, Jessica

,… Petersen, Steven

(2011). Functional Network Organization of the Human Brain. Neuron, 72(4), 665–678. doi:10.1016/j.neuron.2011.09.006

62.

Power, J.D., Mitra, A., Laumann, T.O., Snyder, A.Z., Schlaggar, B.L., & Petersen, S.E. (2014). Methods to detect, characterize, and remove motion artifact in resting state fMRI. NeuroImage, 84, 320–341. doi:10.1016/j.neuroimage.2013.08.048

63.

Riccio, C.A., Reynolds, C.R., Lowe, P., & Moore, J.J. (2002). The continuous performance test: A window on the neural substrates for attention?Archives of Clinical Neuropsychology, 17(3), 235–272. doi:10.1093/arclin/17.3.235

64.

Rodriguez, A.D., Worrall, L., Brown, K., Grohn, B., McKinnon, E., Pearson, C.,… Copland, D.A. (2013). Aphasia LIFT: Exploratory investigation of an intensive comprehensive aphasia programme. Aphasiology, 27(11), 1339–1361. doi:10.1080/02687038.2013.825759

65.

Rose, M.L., Cherney, L.R., & Worrall, L.E. (2013). Intensive Comprehensive Aphasia Programs: An International Survey of Practice. Topics in Stroke Rehabilitation, 20(5), 379–387. doi:10.1310/tsr2005-379

66.

Smith, S.M., Jenkinson, M., Woolrich, M.W., Beckmann, C.F., Behrens, T.E.J., Johansen-Berg, H.,… Matthews, P.M. (2004). Advances in functional and structural MR image analysis and implementation as FSL. NeuroImage, 23, S208–S219. doi:10.1016/j.neuroimage.2004.07.051

67.

Thompson, C.K., Shapiro, L.P., Kiran, S., & Sobecks, J. (2003). The Role of Syntactic Complexity in Treatment of Sentence Deficits in Agrammatic Aphasia. Journal of Speech Language and Hearing Research, 46(3), 591. doi:10.1044/1092-4388(2003/047)

68.

Tomasi, D., & Volkow, N.D. (2012). Resting functional connectivity of language networks: Characterization and reproducibility. Molecular Psychiatry, 17(8), 841–854. doi:10.1038/mp.2011.177

69.

Turken, U., & Dronkers, N.F. (2011). The Neural Architecture of the Language Comprehension Network: Converging Evidence from Lesion and Connectivity Analyses. Frontiers in System Neuroscience, 5. doi:10.3389/fnsys.2011.00001

70.

Ulm, L., Copland, D., & Meinzer, M. (2016). A new era of systems neuroscience in aphasia?Aphasiology, 1–23. doi:10.1080/02687038.2016.1227425

71.

van Hees, S., McMahon, K., Angwin, A., de Zubicaray, G., Read, S., & Copland, D.A. (2014). A functional MRI study of the relationship between naming treatment outcomes and resting state functional connectivity in post-stroke aphasia. Human Brain Mapping, 35(8), 3919–3931. doi:10.1002/hbm.22448

72.

Wallace, S.J., Worrall, L., Rose, T., Le Dorze, G., Cruice, M., Isaksen, J.,… Gauvreau, C.A. (2016). Which outcomes are most important to people with aphasia and their families? an international nominal group technique study framed within the ICF. Disability and Rehabilitation, 39(14), 1364–1379. doi:10.1080/09638288.2016.1194899

73.

Watila, M.M., & Balarabe, S.A. (2015). Factors predicting post-stroke aphasia recovery. Journal of the Neurological Sciences, 352(1-2), 12–18. doi:10.1016/j.jns.2015.03.020

74.

Winans-Mitrik, R.L., Hula, W.D., Dickey, M.W., Schumacher, J.G., Swoyer, B., & Doyle, P.J. (2014). Description of an Intensive Residential Aphasia Treatment Program: Rationale, Clinical Processes, and Outcomes. American Journal of Speech-Language Pathology, 23(2), S330. doi:10.1044/2014_ajslp-13-0102