Advanced Magnetic Resonance Neuroimaging of Language Function Recovery After Aphasic Stroke: A Technical Review

Article (PDF Available)inArchives of physical medicine and rehabilitation 93(1 Suppl):S4-14 · January 2012with74 Reads
DOI: 10.1016/j.apmr.2011.02.023 · Source: PubMed
Abstract
Two advanced magnetic resonance neuroimaging techniques, functional magnetic resonance imaging (fMRI) and diffusion tensor imaging (DTI), have recently made their way into clinically oriented research and hold great promise to study the brain's adaptive changes of function and structure after aphasic stroke, respectively. Such functional and structural neuroplasticity is thought to underlie the recovery of language function, occurring spontaneously and/or in the context of therapeutic intervention. With fMRI, brain activity can be visualized. Spontaneous brain activity, present in multiple brain networks, is measured with resting-state fMRI and language-related brain activity by having the subject perform a language task during scanning (task-based fMRI). With DTI the major white matter tracts, such as the dorsal and ventral language pathways and the commissural fibers, can be visualized and quantified. Both techniques are entirely noninvasive and thus offer the unique opportunity to perform multiple assessments within the same subject. To gain more insight in functional and structural neuroplasticity after aphasic stroke, advanced magnetic resonance neuroimaging studies in specific patient populations, at several stages after stroke and in the course of language recovery, are needed. Such studies will help to clarify the influence of the many factors that play a role in the recovery of language function and are thus vital to further the development of aphasia therapy. Application of these techniques in aphasic stroke patients, however, is not without challenge. The purpose of this article is to discuss the methodologic challenges of fMRI and DTI in the assessment of language recovery after aphasic stroke.
This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
Author's personal copy
SPECIAL COMMUNICATION
Advanced Magnetic Resonance Neuroimaging of Language
Function Recovery After Aphasic Stroke: A Technical Review
Marion Smits, MD, PhD, Evy G. Visch-Brink, MA, PhD, Mieke E. van de Sandt-Koenderman, MA, PhD,
Aad van der Lugt, MD, PhD
ABSTRACT. Smits M, Visch-Brink EG, van de Sandt-
Koenderman ME, van der Lugt A. Advanced magnetic reso-
nance neuroimaging of language function recovery after apha-
sic stroke: a technical review. Arch Phys Med Rehabil 2012;93
(1 Suppl 1):S4-14.
Two advanced magnetic resonance neuroimaging techniques,
functional magnetic resonance imaging (fMRI) and diffusion ten-
sor imaging (DTI), have recently made their way into clinically
oriented research and hold great promise to study the brain’s
adaptive changes of function and structure after aphasic stroke,
respectively. Such functional and structural neuroplasticity is
thought to underlie the recovery of language function, occur-
ring spontaneously and/or in the context of therapeutic in-
tervention. With fMRI, brain activity can be visualized.
Spontaneous brain activity, present in multiple brain net-
works, is measured with resting-state fMRI and language-
related brain activity by having the subject perform a lan-
guage task during scanning (task-based fMRI). With DTI the
major white matter tracts, such as the dorsal and ventral
language pathways and the commissural fibers, can be vi-
sualized and quantified. Both techniques are entirely nonin-
vasive and thus offer the unique opportunity to perform
multiple assessments within the same subject. To gain more
insight in functional and structural neuroplasticity after
aphasic stroke, advanced magnetic resonance neuroimaging
studies in specific patient populations, at several stages after
stroke and in the course of language recovery, are needed.
Such studies will help to clarify the influence of the many
factors that play a role in the recovery of language function
and are thus vital to further the development of aphasia
therapy. Application of these techniques in aphasic stroke
patients, however, is not without challenge. The purpose of
this article is to discuss the methodologic challenges of
fMRI and DTI in the assessment of language recovery after
aphasic stroke.
Key Words: Aphasia; Diffusion tensor imaging; Language
therapy; Magnetic resonance imaging; Rehabilitation; Stroke.
© 2012 by the American Congress of Rehabilitation
Medicine
A
PHASIA IS PRESENT in about 25% to 30% of people
who suffered a stroke.
1,2
The majority of patients receive
some form of language therapy, generally during the first year
postonset. Although the effect of several language therapies
has been shown in small case studies and even in some ran-
domized controlled trials (RCTs), their effectiveness remains
to be established in large scale RCTs.
3-5
While behavioral
studies are used to assess the effect of language therapy on
recovery of language function, current advanced magnetic res-
onance (MR) neuroimaging techniques such as functional mag-
netic resonance imaging (fMRI) and diffusion tensor imaging
(DTI) offer the possibility to study the process of recovery on
a neurophysiologic level, assessing the plasticity of the nervous
system in the recovery process.
It is now well-established that even the adult brain is able to
change in order to adapt to and compensate for functional
deficits. Such neuroplasticity may take many forms, such as the
modification of brain areas that are either directly or indirectly
related to language and changes in the function or number of
synapses (synaptic plasticity) and axonal sprouting leading to
new connections.
6,7
The most direct evidence for neuroplastic-
ity in language function recovery comes from a positron emis-
sion tomography (PET) study by Musso et al,
8
demonstrating
reorganization and reconnection during in-scanner aphasia
therapy.
The debate of language recovery after aphasic stroke centers
on the role of the right hemisphere. In right-handers and most
left-handers, aphasia occurs with stroke in the left hemisphere
and it is postulated that the nondominant right hemisphere
plays an important role in language recovery. One hypothesis is
that homologous language areas in the right hemisphere take
over the role of the damaged language areas in the left hemi-
sphere. Neuroimaging studies of aphasic patients provide some
support for right hemisphere involvement in language recov-
ery.
9
Some studies, however, have interpreted right hemisphere
activation as a reflection of a maladaptive process, related to
poor recovery of language function.
10
Time postonset may be
a relevant factor, with a different role for the right hemisphere
in the postacute stage and in the chronic stage.
11
From the Departments of Radiology (Smits, van der Lugt) and Neurology (Visch-
Brink), Erasmus MC University Medical Centre Rotterdam, Rotterdam; and the
Department of Neurorehabilitation Research, Rehabilitation Centre Rijndam, Rotter-
dam (van de Sandt-Koenderman), The Netherlands.
Presented in part in the form of an educational exhibit to the Radiological Society
of North America, Chicago, IL (exhibit no. 2390CE-e) November 27–December 2,
2005, which was subsequently published in RadioGraphics.
No commercial party having a direct financial interest in the results of the research
supporting this article has or will confer a benefit on the authors or on any organi-
zation with which the authors are associated.
Correspondence to Marion Smits, MD, PhD, Department of Radiology (Hs-224),
Erasmus MC University Medical Centre Rotterdam, PO Box 2040, 3000 CA
Rotterdam, The Netherlands, e-mail: marion.smits@erasmusmc.nl. Reprints will not
be available from the author.
0003-9993/12/9301S-01154$36.00/0
doi:10.1016/j.apmr.2011.02.023
List of Abbreviations
BOLD blood-oxygenation-level-dependent
DTI diffusion tensor imaging
DT-t diffusion tensor tractography
DWI diffusion weighted imaging
FA fractional anisotropy
fMRI functional magnetic resonance imaging
HRF hemodynamic response function
MD mean diffusivity
MR magnetic resonance
PET positron emission tomography
RCT randomized controlled trial
SLF superior longitudinal fasciculus
SNR signal-to-noise ratio
S4
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
Despite the considerable number of patients affected each
year, the large impact on health care and society, and an
increasing interest in aphasia therapy, studies examining the
neural bases of either spontaneous or treatment-induced recov-
ery are limited in number, poorly controlled, and nonuni-
form.
12,13
Also, very little is known about the time-course of
changes in activation patterns in the recovery after stroke.
14
A
comprehensive overview of recent imaging findings is pro-
vided elsewhere in this issue.
15
To gain more insight in the
reorganizational processes, advanced MR neuroimaging stud-
ies in specific patient populations, at several stages after stroke
and in the course of language recovery, are needed. Such
studies will help to clarify the influence of the many factors that
play a role in the recovery of language function and are thus
vital to further the development of aphasia therapy.
9
The 2 advanced MR neuroimaging techniques we focus on
are fMRI and DTI. With fMRI brain activity is measured. With
DTI the microstructure of the white matter can be assessed.
While these techniques are now widely used both in a research
setting and in clinical practice, their application in aphasic
stroke patients is particularly challenging. Careful consider-
ation as well as a thorough understanding of the methodologic
limitations and pitfalls is needed to successfully study neuro-
plasticity of language recovery in such patients. The purpose of
this article is to discuss the methodologic challenges of fMRI
and DTI in the assessment of language recovery after aphasic
stroke. Where possible, we give recommendations based on
recent literature and our own experience.
FUNCTIONAL MR IMAGING
Blood-oxygenation-level-dependent (BOLD) fMRI is now
the most commonly used functional neuroimaging technique to
study the cerebral representation of language processing.
BOLD fMRI takes advantage of the tight link between local
neuronal activity and blood flow (neurovascular coupling).
16,17
When neuronal activity increases locally, local blood flow also
increases, leading to an increase in oxygenated blood that is
disproportionate to the increased need of oxygen for neuronal
activity. The resulting relative decrease of paramagnetic deox-
ygenated hemoglobin leads to an increase of the MR signal in
those areas of the brain that are active.
18,19
Spontaneous brain
activity, present in multiple networks in the brain, is measured
with resting-state fMRI.
20
Such information is of particular
interest in the context of the postulated maladaptive–increased–
activity of the right hemisphere. Language-related brain activ-
ity is measured by having the patient perform a language task
during scanning (task-based fMRI).
There are several factors that make fMRI in aphasic stroke
patients problematic.
21
First, in task-based fMRI, active patient
participation in the performance of the task during scanning is
required, which may be difficult to achieve in the presence of
communicative disabilities. Additional motor disabilities,
which are often present, pose a further difficulty in performing
a task for which motor input (eg, button presses) may be
required. Careful task selection and design, as well as patient
preparation and training are of the utmost importance. Second,
BOLD fMRI relies on the tight link between neuronal activity
and hemodynamic changes, which may be altered in neurovas-
cular disease such as stroke. Common assumptions underlying
task design, data acquisition and analysis may be rendered
invalid due to neurovascular changes and need to be adjusted.
Third, when longitudinally assessing fMRI activation in the
course of language recovery, reproducibility of findings needs
to be established. Finally, there are a wide variety of stroke
induced brain lesions, rendering the conventional pooling of
data for group analyses virtually impossible. Alternative ways
of data analysis need to be considered.
In the sections below, we will first discuss some general
principles of fMRI task design and will subsequently address
each of these issues, giving a summary and recommendations
in the final section.
Paradigm Design: General Principles
Because signal changes are small and occur at a delay of
approximately 6 seconds, careful design of the task that is
performed by the subject during scanning is required. A task or
so-called paradigm typically consists of active and control
conditions. A rough distinction can be made between para-
digms that are blocked and those that are event-related.
22
Blocked paradigms consist of a sequence of blocks of an active
or control condition, each commonly lasting 20 to 40 seconds.
Within each block, a series of trial events of 1 condition is
presented and the signal acquired during 1 block is then com-
pared with the other block(s) constituting a different condition.
Blocked paradigms are statistically robust, because a lot of
signal is acquired for each condition, but are restrained, leaving
not much room for unexpected or short stimuli, and subjects
may develop cognitive strategies for responding to items within
a block. Short, (pseudo) random stimulus presentation is pos-
sible within an event-related design, during which individual
trial events, each representing a specific condition, are pre-
sented in random order and rapid succession. An event-related
design therefore offers the possibility to present unexpected
stimuli as well as many different conditions, rendering it very
flexible, but statistically less robust because the signal that is
acquired per condition is generally low.
Task Difficulty
In studying patients with a neurologic and/or cognitive def-
icit, it is crucial that the task difficulty is adapted to meet the
patient’s ability to perform the task. A task that is too difficult
will result in underperformance or dropout, resulting in de-
creased or absent activation. Brain activity, however, will also
be low if no challenge is posed. Generally, we aim for a task
performance level between 70% and 90%, as is established
during pilot studies. In a relatively homogeneous patient pop-
ulation, one can use a task with a fixed level of difficulty that
all patients are expected to meet. If patients are not too badly
affected, the same task can be used both for patients and
healthy controls. To accommodate heterogeneity in the severity
of disability, one can use a task with several levels of difficulty
in a so-called parametric design. Alternatively, one can test the
patient’s performance prior to scanning and adjust task diffi-
culty on an individual basis. Although this introduces a certain
amount of heterogeneity with respect to the stimulus para-
digms, overall task performance is expected to be the same.
Training beforehand is important to ensure adequate task per-
formance, while overlearning should be avoided.
Stimulus Paradigms
A multitude of stimulus paradigms has been developed,
published, and implemented for language stimulation. Well-
known tasks are verbal fluency and passive listening para-
digms, as shown in detail in table 1.
11,14,23-49
Verbal fluency
paradigms require the selection and production of words ac-
cording to a given concept (semantic fluency) or phoneme
(phonologic fluency), involving language expression and exec-
utive function. By including several levels of stimulus com-
plexity, a parametric design can be created. Verbal fluency
tasks give rise to activation in the classic Broca’s (routinely)
S5
MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
Table 1: Commonly Used fMRI Language Paradigms for the Assessment of Global Language Function and Language Components
Paradigm Instruction Presentation Response Comments
Global language function
Verbal fluency Generate a verb from a given noun
or picture.
24-28
Auditory or visual Covert* or overt
Visual presentation introduces
additional processes and
difficulty especially for
aphasic patients.
Generate a single word starting
with a given letter.
29
Auditory or visual Covert* or overt
If too difficult, subject may be
instructed to produce the
next letter in the alphabet
instead.
Complete the word with a given
stem.
30
Visual Covert* or overt
...
Generate a single word from a
given category.
29,31,32
Auditory or visual Covert* or overt
...
Passive listening Listen to standard
text/story/sentences.
28,29,33,34
Auditory . . . Easy to perform (also in
severely aphasic patients)
and implement.
Listen to text from the subject’s
favorite novel/magazine.
Auditory . . . Very useful in small children,
even when subject is asleep
or sedated.
Read text or sentences, followed
by a question.
35
Visual Covert* or overt
Involves comprehension.
Picture naming Name the presented picture of a
common object.
36-38
Visual Covert* or overt
Involves all language
components on word level.
Generate a sentence describing the
presented picture.
39
Visual Covert* or overt
Involves all language
components.
Phonologic processing
Rhyme judgment Decide whether presented word
pairs rhyme.
40-42
Auditory or visual Button press Visual presentation introduces
additional difficulty
(grapheme to phoneme
conversion) and is only
possible with irregularly
spelled homophones.
Decide whether the presented
picture rhymes with the
simultaneously presented
word.
14,36
(Auditory and)
visual
Button press Visual presentation introduces
additional difficulty.
Lexical decision Decide whether word/sentence is
phonologically correct.
43
Auditory or visual Button press Visual presentation introduces
additional difficulty (grapheme
to phoneme conversion).
Decide whether presented word
pairs are homophones.
44
Visual Button press Requires grapheme to
phoneme conversion.
Letter fluency Name as many nouns starting with
a given letter.
30,37,41,45
Auditory or visual Covert* or overt
Also involves executive function.
Semantic processing
Semantic decision Decide whether presented word
pairs belong to the same
semantic category.
40
Auditory or visual Button press . . .
Decide whether presented word
pairs are synonyms.
44,46
Auditory or visual Button press . . .
Decide whether the presented
word belongs to a given
category.
26,37,42,43,47,48
Auditory or visual Button press . . .
Sentence judgment Decide whether the presented
sentence is semantically
correct.
11,49
Auditory or visual Button press . . .
Picture to word/
sentence matching
Decide whether given word/
sentence meaning corresponds
with presented picture.
14,26
Visual Button press May be (too) difficult for
aphasic patient.
Category fluency Name as many nouns belonging to
a given category.
39,45
Auditory or visual Overt
Also involves executive function.
*In case of covert responses only, task performance is not monitored.
Consider sparse sampling acquisition to reduce motion artifact.
S6 MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
and Wernicke’s (often) areas in the dominant hemisphere, as
well as in the premotor cortex, posterior fusiform gyrus, middle
temporal gyrus, dorsolateral prefrontal cortex, supplementary
motor area, and anterior cingulate gyrus.
23
Tasks concerning
passive listening consistently give rise to activation in the
classic Wernicke’s area, but commonly also in the expressive
speech areas in the inferior frontal gyrus in the dominant
hemisphere. The latter finding may be due to the subject’s
covertly rehearsing the heard text with coarticulation.
Both verbal fluency and passive listening paradigms give
rise to robust activation of large parts of the language network
and are excellent tasks to get an overall measure of language
activation in the brain. An advantage of the passive listening
task is that it does not require active participation from the
subject, and is therefore suitable for use in even severely
aphasic patients or small children.
33,50
A certain degree of
parametricity can be introduced by using blocks of text with
varying levels of complexity with respect to, for example
sentence structure or the use of less or more commonly used
words.
For studying neuroplasticity of language recovery with lan-
guage therapy, however, we will also want to disentangle the
several components of language, because it seems that the
areas of the brain engaged to support recovery may differ
depending on the type of treatment provided, as through treat-
ment separate language components may be targeted or
trained.
9
Furthermore, when specific types of aphasia or disor-
ders on linguistic levels are studied, we will specifically want
to monitor the recovery and reorganizational processes in-
volved in the affected language component.
The design of tasks to isolate the components of language
processing is much more complex, as the components of lan-
guage are thoroughly intertwined, and careful stimulus selec-
tion and contrast with a closely matched control condition is
required to separate 1 language component from the others.
The most commonly studied language components with fMRI
are phonologic and semantic processing, assessed with phono-
logic paradigms such as letter decision, word completion, pho-
nologic fluency, and rhyming and semantic paradigms such as
category decision, semantic fluency, semantic association, and
picture to text matching, respectively. More detail on these, and
other commonly used tasks, can be found in table 1, as well as
in the excellent reviews by Price,
51
McGraw and colleagues,
23
.
However well designed the task though, a certain degree of
contribution from other language components than the one of
interest is inevitable (fig 1).
In addition to stimulus selection, several other paradigm
design decisions need to be made. First, we need to decide
whether to present the stimuli auditorily or visually. Visual
stimulus presentation offers the advantage that stimuli are not
degraded by the loud scanner background noise, as is the case
with auditorily presented stimuli. Also, hearing deficits, com-
monly present in the elderly population, are not an issue, while
visual acuity deficits may be easily overcome with MR com-
patible glasses. The disadvantage, however, is that additional
levels of processing are introduced that we may not necessarily
be interested in and which may present an unnecessary addi-
tional challenge for aphasic patients. This is especially true for
phonologic processing, for which the processing of visually
presented phonologically oriented stimuli requires grapheme to
phoneme conversion.
Second, we need to decide whether the task is performed
overtly (spoken) or covertly (silent), and, relatedly, how in-
scanner task performance is monitored. It is highly recom-
mended to obtain in-scanner performance data not only to
assure that the task is performed correctly, but also to be able
to account for differences in task performance in post hoc
analyses. With overt task performance response, monitoring is
relatively easy as overt responses can be registered and re-
corded. With covert task performance, response monitoring can
only be recorded indirectly, for instance by means of response
buttons in judgment-oriented paradigms. Care needs to be
taken that the additional activity related to responding but
unrelated to language processing (eg, finger movement by
pressing the button) is balanced out in the control condition.
Furthermore, additional motor deficits that may be present need
also be considered, for example by having the task consistently
be performed with the nondominant (and presumably nonaf-
fected) hand.
While overt task performance may thus seem the obvious
choice from a performance-monitoring point of view, several
disadvantages render it less optimal than covert task perfor-
mance. These are predominantly due to head motion and the
movement of air in the paranasal sinuses and nasal and oral
cavity during speech, the latter leading to unpredictable sus-
ceptibility and distortion artifacts. Head motion artifacts are not
easily compensated for, as they are inherently response related
and thus may severely reduce statistical power.
52
With training,
excessive head motion may be reduced, but is rarely mini-
mized. Sparse sampling acquisition techniques (see below) or a
hemodynamic delay design
53
are better suited to allow for overt
task performance than conventional continuous scanning, but
come at the cost of lower sensitivity and longer acquisition
times and are therefore best reserved for those instances that
overt responses are considered an absolute requirement.
Data Acquisition: Continuous Versus Sparse Sampling
Acquisition
When using auditory stimuli, fMRI is particularly challeng-
ing due to an interaction between the experimental auditory
stimuli and the extremely loud background scanner noise.
54-57
As well as being very loud (up to 110dB), the MR imaging
scanner sound is an amplitude-modulated periodic sound with
a complex spectrum that very likely interacts with the experi-
mentally delivered stimuli.
57,58
Subjects may also be engaged
in processes different from auditory perception, because they
have to extract the stimulus from the background MR-gener-
ated noise and will supposedly need to recruit more areas in the
brain than strictly necessary for performing the task (fig 2).
This issue is most prominent when phonologic processing is
studied. With standard MR compatible headphones, some at-
tenuation (up to 30dB) of the background scanner noise can be
achieved. Silent fMRI techniques, such as the BURST se-
quence, are very effective in reducing acoustic noise,
59
but
most tend to be too slow for fMRI studies.
60
Longer noise-free
periods during acquisition— known as sparse temporal sam-
pling—are also useful in reducing the amount of scanner-
generated noise. Such sparse sampling techniques take advan-
tage of the fact that the hemodynamic response to the increase
in neuronal activity is delayed. It is therefore possible to
acquire data at a delay after stimulus presentation, while the
auditorily presented stimuli are not degraded by the scanner
noise. They have been shown to improve fMRI activation of
the auditory and language systems, but the amount of informa-
tion acquired is usually decreased and acquisition times are
long.
56,58,61,62
A compromise can be found in the clustered
volume acquisition technique.
63-65
This method has the advan-
tage of a global increase in efficiency, while retaining sufficient
silent gaps during which the subject can clearly perceive the
auditory stimulus. We previously used this technique success-
fully in assessing both phonologic and auditory processing,
allowing for detection of fMRI activation in both cortical and
S7
MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
subcortical structures without the need for lengthy acquisition
times.
40,65,66
Compared with continuous scanning, however, acquisition
time with clustered volume acquisition is still at least doubled.
Longer scanning times introduce the risk of motion artifacts,
patient motivational and attentional issues, and a compromise
in the number of language components that can be studied
within a scanning session of a given duration. Additionally, and
possibly more importantly, sparse sampling techniques heavily
rely on the hemodynamic delay after neuronal activity, with
data being acquired at the assumed peak of the BOLD re-
sponse. As discussed below, assumptions about the hemody-
namic delay after neuronal activity, expressed as the hemody-
namic response function (HRF), and related BOLD responses
may not be valid in patients who suffered a stroke, which may
severely reduce BOLD sensitivity in sparse sampling acquisi-
tion techniques.
Neurovascular Changes
One of the major issues with fMRI in stroke patients is the
fact that BOLD fMRI relies on the tight link between neuronal
activity and hemodynamic changes. Neuronal activity is only
measured indirectly, assumed to underlie the measured changes
in blood oxygenation, which in turn are assumed to be the
result of hemodynamic changes in response to neuronal activ-
ity. While these assumptions may be justified in healthy, young
volunteers, they may be utterly invalid in the elderly patient
population with neurovascular disease, such as the majority of
stroke patients.
The first problem arises from the assumption that the hemo-
dynamic response after an increase in neuronal activity peaks at
approximately 6 seconds. Typically, the stimulus paradigm is
convolved with this HRF prior to general linear model analysis.
Studies of stroke patients, however, showed that the HRF may
actually peak up to 20 seconds after stimulus presentation.
67,68
Fig 1. Three dimensional brain renderings with superposition of activation maps (thresholded at T>3.5 and cluster size>10 voxels) from
fixed effects group analyses of 6 healthy, right-handed subjects.
40
(A) Activation during an auditorily presented rhyme judgment task; (B)
activation during an auditorily presented semantic decision task. During both tasks, activation is seen bilaterally and left-lateralized in the
posterior temporal regions (classic Wernicke’s area). Exclusively left-sided inferior frontal gyrus activation (classic Broca’s area) is predom-
inantly seen during the rhyme judgment task.
S8 MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
Such findings are, at least in part, thought to be the result of
underlying age- and disease-related changes in the vascular
bed, such as increased vascular tortuosity resulting in differ-
ences in the spatial organization of intracerebral arterioles,
capillaries, and venules.
69
This is not too problematic when a
blocked stimulus paradigm is used, which is not very sensitive
to temporal discrepancies between the modeled stimulus par-
adigm and measured signal changes, but it becomes a real issue
when an event-related design is used, which heavily relies on
the exact modeling of stimulus and response timing convolved
with the HRF.
Second, in the very early stages (first week) after stroke, the
response of the local vasculature to ischemia is to dilate max-
imally, disrupting the normal neurovascular coupling response
and resulting in a reduced or absent BOLD signal.
68,70
For this
and practical reasons, BOLD fMRI is not recommended in the
first week after stroke.
Third, hemodynamic compromise may exist beyond local
changes in neurovascular response. Cerebrovascular stenotic or
occlusive disease, a common cause of ischemic stroke, may
result in a hemodynamic compromise more extensive than in
the ischemic brain tissue alone. Resulting compensatory hemo-
dynamic changes consisting of maximal vasodilatation in the
affected brain region(s) may lead to an inability of the local
vasculature to respond to neuronal activity and consequently to
reduced or absent BOLD sensitivity and false negative re-
sults.
24,67,69
BOLD sensitivity may roughly be assessed with
breath hold fMRI,
26,71
while perfusion studies combined with a
physiologic or pharmacologic challenge provide direct mea-
sures of the hemodynamic reserve.
72
Reproducibility
When assessing changes in language activation as a function
of spontaneous recovery and/or language therapy, the effect
size needs to be greater than the variability between 2 scans, or
the intraindividual variability. In a group of unselected patients
with drug-resistant focal epilepsy, Fernández et al
46
showed
high intraindividual reproducibility across the hemispheres,
resulting in highly reliable laterality indices, but not within
each of the hemispheres, with significant variability in the
location of particularly the temporoparietal regions. In stroke
patients, Chen and Small
73
found that language activation was
less reliable than in healthy controls, further affirming the need
to assess and account for intraindividual variation of activation
patterns measured with fMRI.
Analysis
Changes measured with fMRI are small, and, when subtle
effects are studied, may not even be detectable on an indi-
vidual level. Typically, data from multiple subjects are
combined in group analyses to increase statistical power and
to make inferences on a population level. Given the large
heterogeneity in stroke patients regarding lesion size, shape,
and location, conventional group analyses are problematic.
The perilesional region, for instance, will vary from patient
Fig 2. Axial T1-weighted sections of the brain with superposition of activation maps (thresholded at T>5 and cluster size>10 voxels) from
fixed effects group analyses of 6 healthy, right-handed subjects during the performance of an auditorily presented semantic decision task.
40
With continuous acquisition (left-hand figure), activation is much more widespread, with additional activation in the frontal language areas,
compared with clustered volume acquisition (right-hand figure). Activation is expected to be lateralized to the left hemisphere in these
right-handers, which is evident with clustered volume acquisition, but less so with continuous acquisition. Presumably, because stimuli need
to be extracted from the background scanner noise with continuous acquisition, additional brain areas, such as for secondary auditory
processing and phonologic processing, are recruited. Abbreviations: L, left; R, right.
S9MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
to patient and will thus need to be defined on an individual
level. The presence of a lesion also hinders the normaliza-
tion of the patients’ brains to a standard brain template.
Using unified segmentation methods as well as masking the
lesion prior to normalization may be necessary to improve
interindividual alignment of analogous brain areas and to
avoid underestimation of the lesion.
74,75
The analysis of fMRI data of stroke patients may thus often
rely on a careful study of individuals rather than on, or in
addition to, the study of a group of patients.
9
Group analyses
will need to be based on region of interest analyses, in which
the known language areas and their homologues in the right
hemisphere are identified in the individual patient. It is advis-
able to include a neurologically intact, age- and sex-matched
control group to identify task-specific language networks that
may be used to define the study-specific regions of interest.
Perilesional areas may be identified on an individual patient
level as additional regions of interest. Further regions of inter-
est may be considered based on the supposedly underlying
neurophysiologic processes of the aphasia therapy under study.
To account for heterogeneity in severity of aphasia, degree of
recovery, and task performance, behavioral (performance) data
need to be collected and used in the analyses.
More advanced analyses techniques than the commonly used
model-based analyses are model-free and functional connec-
tivity analyses.
76
Model-free analyses are data rather than
model driven, so no assumptions need to be made regarding the
underlying stimulus paradigm and timing of the HRF, render-
ing these an interesting alternative for the analysis of stroke
patients with unpredictable HRF. Functional connectivity anal-
yses identify brain regions that are simultaneously active while
spatially remote, thus implying a functional connection be-
tween these regions, although a causal relationship cannot be
established.
35,77-80
Functional connectivity analyses allow for a
more integral assessment of the language network, as shown in
recent fMRI and PET studies.
81,82
In the latter study, aphasic
patients demonstrated selective disruption of the normal func-
tional connection between the left and right anterolateral supe-
rior temporal cortices, the degree of which correlated nega-
tively with the degree of recovery.
Summary and Recommendations
It is clear that fMRI of language in stroke patients is not
without challenge and that careful consideration of study and
task design, data acquisition, and data analysis needs to be
taken. There is no standard approach, as deciding factors de-
pend on the aim of the study, the patient population, and
logistical and practical issues. Below we give some general
recommendations.
Given the need for robust and strong activation on an indi-
vidual level, as well as uncertainties regarding the HRF, a
blocked paradigm design is preferred owing to its relatively
large statistical power and insensitivity to temporal discrepan-
cies between the stimulus paradigm and the measured BOLD
response. If time permits, we would suggest including both a
robust, global language task as well as tasks aimed at the
separate components of language. These are of particular in-
terest in the assessment of the effect of language therapy, for
which the fMRI task ideally closely matches the therapy.
Covert task performance is preferred to overt task perfor-
mance, due to difficulties related to head motion, distortion,
and susceptibility artifacts. An alternative measure of perfor-
mance monitoring, however, needs to be sought to assure
ourselves of adequate task performance and for the collection
of in-scanner behavioral data. If overt task performance is
required, for instance to specifically study language production,
sparse sampling techniques, or hemodynamic delay, modeling
should be considered.
If sparse sampling techniques are used, special consideration
needs to be given to the fact that the underlying assumptions
regarding the timing and shape of the HRF may be invalid in
stroke patients. Prior assessment of the HRF may be needed to
adapt the scanning sequence accordingly. Individual assess-
ment of the HRF may also be required when an event-related
task design is used, for which exact timing information is
critical. If the HRF is not known, uncertainties regarding the
HRF can be taken into account by modeling the canonical HRF
with time and dispersion derivatives.
Neurovascular changes may further be taken into account by
assessing carotid and intracranial artery stenotic disease using
computed tomography or MR angiography, which are com-
monly performed in the diagnostic work-up after stroke. Fur-
ther information on— compromised—perfusion can now be
obtained quantitatively and noninvasively with arterial spin
labeling.
83
Ideally, intraindividual variability is assessed by performing
multiple scanning sessions in each patient pre- and posttherapy.
This may, however, logistically not be possible and result in
undesirable patient dropout. If robust tasks are used and effects
are expected to be relatively large, intraindividual variability
will be less of an issue and test-retest assessment may not be
needed. However, if effects are expected to be subtle, perform-
ing 2 scanning sessions before treatment may be a minimal
requirement.
9
Finally, advanced analysis techniques, such as model free
analyses and functional connectivity analyses, should be con-
sidered as alternatives of or additions to standard model-based
analyses to further our understanding of the functional organi-
zation of the language network, its disruption in aphasia, and
reorganization in language function recovery.
35,84,85
DIFFUSION TENSOR IMAGING
In addition to functional information obtained with fMRI,
information on structural connections within the brain can
be obtained with DTI. Diffusion weighted imaging (DWI)
provides image contrast sensitive to differences in the dif-
fusion of water molecules.
86-88
In DTI, diffusion weighted
gradients are applied in multiple directions, such that the
anisotropy of diffusion can be assessed in addition to mean
diffusivity (MD). A higher anisotropy of diffusion reflects
the motion of water molecules favored in a specific direc-
tion, such as parallel to highly structured white matter fibers.
This information can then be translated into a vector field.
When vectors that have the same orientation are combined,
the course of white matter tracts may be visualized, which is
known as diffusion tensor tractography (DT-t).
89-91
With
such fiber tracking techniques, connections between differ-
ent areas of the brain can be depicted and quantified in terms
of MD and fractional anisotropy (FA).
Diffusion Tensor Tractography
DT-t approaches can be roughly divided in deterministic and
probabilistic fiber tracking techniques. It is important to realize
that both techniques are generally based on the same line propa-
gation technique in which a tracking line is propagated from a
starting or seed point.
92,93
In deterministic tractography, only 1
trajectory is generated from a given seed point. In probabilistic
tractography, on the other hand, a large number of trajectories are
propagated from the same seed point. For each trajectory, the
number of times this same trajectory was generated is displayed,
thus generating a map of all probable trajectories as well as a
S10
MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
measure of probability for each trajectory. While probabilistic
tractography is much more computationally demanding than de-
terministic tractography, it is a powerful approach to explore
possible connections between brain regions, and to follow fiber
tracts when they course into the grey matter.
94
The successful
application of probabilistic tractography has in recent years been
demonstrated by the identification of several white matter struc-
tures interconnecting cortical language areas in addition to the
well-known arcuate fasciculus.
95
Language White Matter Pathways
The arcuate fasciculus (fig 3) is the best-known language-
related white matter pathway, connecting the frontal expressive
language areas (classic Broca’s area) with the posterior temporo-
parietal receptive language areas (classic Wernicke’s area). It is
generally considered a subdivision of the superior longitudinal
fasciculus (SLF), and as such cannot be assessed separately from
the SLF with volumetric white matter analysis but needs to be
isolated with DT-t.
96
In addition to this so-called dorsal language
pathway, a ventral connection between the temporal lobe language
areas and the frontal lobe through the uncinate fasciculus has
recently been proposed.
97,98
In a combined fMRI and DT-t study,
Saur et al
98
were able to confirm the hypothesis of a dual func-
tional system, with the dorsal pathway being involved in sound-
to-motor mapping and the ventral pathway in sound-to-meaning
integration. Catani et al
99
further segregated the dorsal language
stream into 1 long direct segment—directly interconnecting clas-
sical Broca’s and Wernicke’s areas—and 2 short indirect seg-
ments with an additional connection to Geschwind’s territory
(inferior parietal lobule). As well as the association white matter
tracts, commissural fiber tracts, such as the corpus callosum and
the anterior commissure, may also play a role in language function
and recovery. The corpus callosum is the largest white matter
structure in the brain connecting the 2 hemispheres. The anterior
commissure directly connects the left and right anterior temporal
lobes, including the left and right anterolateral superior temporal
sulcus, the functional connectivity of which seems to be strongly
correlated with language recovery.
82
Considerations and Pitfalls
Considerations regarding DTI acquisition are not so much in
the realm of patient cooperation, other than the requirement for the
patient to lie still, but are found in the balance of a signal-to-noise
ratio (SNR), spatial resolution, number of diffusion weighted
gradient directions, and acquisition time.
91,93,100
Motion is an
important source of noise and therefore needs to be minimized,
both by keeping acquisition time within reasonable limits
(15min), and by using cardiac gating to reduce physiologic
noise introduced by the cardiac cycle. Increasing the number of
diffusion weighted gradients increases the accuracy of tensor
estimation, while acquiring low and high b-value images accord-
ingtoa1to8-9ratio ensures optimal SNR.
91,100
Increasing spatial
resolution reduces partial volume effects, resulting in more accu-
rate tensor estimation. For a comprehensive overview of these and
other considerations as well as limitations and pitfalls associated
with DTI, we refer to the recently published comprehensive re-
view by Jones.
93
For statistical analysis of DTI data, FA is the most com-
monly used DTI metric, as it has been shown to map diffusion
anisotropy with the greatest SNR and with the greatest de-
tail.
100
Reduction of FA is generally interpreted as loss of white
matter integrity. This, however, represents 1 of the major
pitfalls in the analysis and interpretation of DTI data. A voxel
contains a multitude of white matter fibers or even several fiber
Fig 3. Axial and coronal T1-weighted sec-
tions of the brain of a single, right-handed
healthy volunteer with superposition of
language activation as well as DT-t of the
arcuate fasciculus, interconnecting the
frontal and posterior temporoparietal lan-
guage areas. Note the leftward lateraliza-
tion of both the language activation and
the arcuate fasciculus.
96
S11MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
tracts. FA in voxels containing fibers that cross, splay, kiss, or
abruptly change direction may be artificially low, due to the
averaging of multiple fiber orientations within such voxels. A
further, and partly related, pitfall in the interpretation of DTI
data is the concept of connectivity. While it is compelling to
interpret high FA of white matter tracts connecting 2 brain
regions as a representation of strong connectivity between such
regions, such an interpretation is not justified given the current
limitations of DTI.
100
CONCLUSIONS
Advanced MR neuroimaging techniques, such as fMRI and
DTI, are powerful tools for the assessment of the neurophysi-
ologic changes underlying language recovery after aphasic
stroke. Such information is not only fundamental for an under-
standing of the recovery process itself, occurring spontane-
ously and/or in response to language therapy, but also vital for
the further development and refinement of language therapies.
Many factors play a role in the recovery of language function,
including lesion characteristics, type and severity of aphasia,
and type, timing, and intensity of aphasia treatment. Future
studies that explicitly consider such patient- and treatment-
related variables will help us to separate the essential from the
nonessential parts of aphasia therapy and henceforth guide
future treatment strategies.
References
1. Engelter ST, Gostynski M, Papa S, et al. Epidemiology of
aphasia attributable to first ischemic stroke: incidence, severity,
fluency, etiology, and thrombolysis. Stroke 2006;37:1379-84.
2. Laska AC, Hellblom A, Murray V, Kahan T, Von Arbin M.
Aphasia in acute stroke and relation to outcome. J Intern Med
2001;249:413-22.
3. Cappa S, Benke T, Clarke S, Rossi B, Stemmer B, Van Heugten
C. EFNS guidelines on cognitive rehabilitation: report of an
EFNS task force. Eur J Neurol 2005;12:665-80.
4. Greener J, Enderby P, Whurr R. Speech and language therapy for
aphasia following stroke. Cochrane Database Syst Rev 2000;2:
CD000425.
5. Kelly H, Brady MC, Enderby P. Speech and language therapy for
aphasia following stroke. Cochrane Database Syst Rev 2010
May 12;(5):CD000425.
6. Rijntjes M, Weiller C. Recovery of motor and language abilities
after stroke: the contribution of functional imaging. Prog Neu-
robiol 2002;66:109-22.
7. Price CJ, Crinion J. The latest on functional imaging studies of
aphasic stroke. Curr Opin Neurol 2005;18:429-34.
8. Musso M, Weiller C, Kiebel S, Müller SP, Bülau P, Rijntjes M.
Training-induced brain plasticity in aphasia. Brain 1999;122:
1781-90.
9. Thompson CK, den Ouden DB. Neuroimaging and recovery of
language in aphasia. Curr Neurol Neurosci Rep 2008;8:475-83.
10. Belin P, Van Eeckhout P, Zilbovicius M, et al. Recovery from
nonfluent aphasia after melodic intonation therapy: a PET study.
Neurology 1996;47:1504-11.
11. Saur D, Lange R, Baumgaertner A, et al. Dynamics of language
reorganization after stroke. Brain 2006;129:1371-84.
12. Doesborgh S, van de Sandt-Koenderman M, Dippel D, van
Harskamp F, Koudstaal P, Visch-Brink E. Effects of semantic
treatment on verbal communication and linguistic processing in
aphasia after stroke: a randomized controlled trial. Stroke 2004;
35:141-6.
13. Thompson C, LaPointe L. Functional neuroimaging: applications
for studying aphasia. Aphasia and related neurogenic language
disorders. New York: Thieme; 2005. p 19-38.
14. Fernandez B, Cardebat D, Demonet JF, et al. Functional MRI
follow-up study of language processes in healthy subjects and
during recovery in a case of aphasia. Stroke 2004;35:2171-6.
15. Saur D, Hartwigsen G. Neurobiology of language recovery after
stroke: lessons from neuroimaging studies. Arch Phys Med Re-
habil 2012;93:S15-25.
16. Gjedde A, Jezzard P, Matthews P, Smith S. Brain energy me-
tabolism and the physiological basis of the haemodynamic re-
sponse. Functional MRI: an introduction to methods. Jezzard P,
Matthews P, Smith S, editors. Oxford: Oxford University Pr;
2002. p 37-66.
17. Ogawa S, Lee T, Kay A, Tank D. Brain magnetic resonance
imaging with contrast dependent on blood oxygenation. Proc
Natl Acad SciUSA1990;87:9868-72.
18. Matthews P, Jezzard P, Matthews P, Smith S. An introduction to
functional magnetic resonance imaging of the brain. Functional
MRI: an introduction to methods. Jezzard P, Matthews P, Smith
S, editors. Oxford: Oxford University Pr; 2002. p 3-34.
19. Thulborn K, Waterton J, Matthews P, Radda G. Oxygenation
dependence of the transverse relaxation time of water protons
in whole blood at high field. Biochim Biophys Acta 1982;
714:265-70.
20. Damoiseaux JS, Rombouts SA, Barkhof F, et al. Consistent
resting-state networks across healthy subjects. Proc Natl Acad
Sci USA 2006;103:13848-53.
21. Crosson B, McGregor K, Gopinath KS, et al. Functional MRI of
language in aphasia: a review of the literature and the method-
ological challenges. Neuropsychol Rev 2007;17:157-77.
22. Donaldson D, Buckner R, Jezzard P, Matthews P, Smith S.
Effective paradigm design. Functional MRI: an introduction to
methods. Jezzard P, Matthews P, Smith S, editors. Oxford:
Oxford University Pr; 2002. p 177-95.
23. McGraw P, Mathews V, Wang Y, Phillips M. Approach to
functional magnetic resonance imaging of language based on
models of language organization. Neuroimaging Clin N Am
2001;11:343-53, x.
24. Smits M, Wieberdink RG, Bakker SL, Dippel DW. Functional
magnetic resonance imaging to determine hemispheric language
dominance prior to carotid endarterectomy. J Neuroimaging
2011;21:e162-5.
25. Cao Y, Vikingstad EM, George KP, Johnson AF, Welch KM.
Cortical language activation in stroke patients recovering from
aphasia with functional MRI. Stroke 1999;30:2331-40.
26. van Oers CA, Vink M, van Zandvoort MJ, et al. Contribution of
the left and right inferior frontal gyrus in recovery from aphasia.
A functional MRI study in stroke patients with preserved hemo-
dynamic responsiveness. NeuroImage 2010;49:885-93.
27. Benson RR, FitzGerald DB, LeSueur LL, et al. Language dom-
inance determined by whole brain functional MRI in patients
with brain lesions. Neurology 1999;52:798-809.
28. Tzourio-Mazoyer N, Josse G, Crivello F, Mazoyer B. Interindi-
vidual variability in the hemispheric organization for speech.
Neuroimage 2004;21:422-35.
29. Gaillard WD, Balsamo L, Xu B, et al. fMRI language task panel
improves determination of language dominance. Neurology
2004;63:1403-8.
30. Yetkin FZ, Hammeke TA, Swanson SJ, et al. A comparison of
functional MR activation patterns during silent and audible lan-
guage tasks. AJNR Am J Neuroradiol 1995;16:1087-92.
31. Peck KK, Moore AB, Crosson BA, et al. Functional magnetic
resonance imaging before and after aphasia therapy: shifts in
hemodynamic time to peak during an overt language task. Stroke
2004;35:554-9.
32. Crosson B, Moore AB, Mcgregor KM, et al. Regional changes in
word-production laterality after a naming treatment designed to
S12 MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
produce a rightward shift in frontal activity. Brain Lang
2009;111:73-85.
33. Crinion JT, Warburton EA, Lambon-Ralph MA, Howard D,
Wise RJ. Listening to narrative speech after aphasic stroke: the
role of the left anterior temporal lobe. Cereb Cortex 2006;16:
1116-25.
34. Schlosser MJ, Aoyagi N, Fulbright RK, Gore JC, McCarthy G.
Functional MRI studies of auditory comprehension. Hum Brain
Mapp 1998;6:1-13.
35. Rykhlevskaia E, Gratton G, Fabiani M. Combining structural and
functional neuroimaging data for studying brain connectivity: a
review. Psychophysiology 2008;45:173-87.
36. Léger A, Démonet JF, Ruff S, et al. Neural substrates of spoken
language rehabilitation in an aphasic patient: an fMRI study.
Neuroimage 2002;17:174-83.
37. Deblaere K, Backes WH, Hofman P, et al. Developing a com-
prehensive presurgical functional MRI protocol for patients with
intractable temporal lobe epilepsy: a pilot study. Neuroradiology
2002;44:667-73.
38. Roux FE, Boulanouar K, Lotterie JA, Mejdoubi M, LeSage JP,
Berry I. Language functional magnetic resonance imaging in
preoperative assessment of language areas: correlation with di-
rect cortical stimulation. Neurosurgery 2003;52:1335-45; discus-
sion 45-7.
39. Stippich C, Rapps N, Dreyhaupt J, et al. Localizing and lateral-
izing language in patients with brain tumors: feasibility of rou-
tine preoperative functional MR imaging in 81 consecutive pa-
tients. Radiology 2007;243:828-36.
40. Smits M, Visch-Brink E, Schraa-Tam CK, Koudstaal PJ, Van
Der Lugt A. Functional MR imaging of language processing: an
overview of easy-to-implement paradigms for patient care and
clinical research. Radiographics 2006;26 Suppl 1:S145-58.
41. Lurito JT, Kareken DA, Lowe MJ, Chen SH, Mathews VP.
Comparison of rhyming and word generation with FMRI. Hum
Brain Mapp 2000;10:99-106.
42. Calvert G, Brammer M, Morris R, Williams S, King N, Mat-
thews P. Using fMRI to study recovery from acquired dysphasia.
Brain Lang 2000;71:391-9.
43. Zahn R, Schwarz M, Huber W. Functional activation studies of
word processing in the recovery from aphasia. J Physiol Paris
2006;99:370-85.
44. Gitelman DR, Nobre AC, Sonty S, Parrish TB, Mesulam MM.
Language network specializations: an analysis with parallel task
designs and functional magnetic resonance imaging. Neuroimage
2005;26:975-85.
45. Meinzer M, Flaisch T, Wilser L, et al. Neural signatures of
semantic and phonemic fluency in young and old adults. J Cogn
Neurosci 2009;21:2007-18.
46. Fernández G, Specht K, Weis S, et al. Intrasubject reproducibil-
ity of presurgical language lateralization and mapping using
fMRI. Neurology 2003;60:969-75.
47. Binder JR, Swanson SJ, Hammeke TA, Sabsevitz DS. A com-
parison of five fMRI protocols for mapping speech comprehen-
sion systems. Epilepsia 2008;49:1980-97.
48. Szaflarski JP, Binder JR, Possing ET, McKiernan KA, Ward BD,
Hammeke TA. Language lateralization in left-handed and ambi-
dextrous people: fMRI data. Neurology 2002;59:238-44.
49. Gartus A, Foki T, Geissler A, Beisteiner R. Improvement of
clinical language localization with an overt semantic and syntac-
tic language functional MR imaging paradigm. AJNR Am J
Neuroradiol 2009;30:1977-85.
50. Crinion J, Price CJ. Right anterior superior temporal activation
predicts auditory sentence comprehension following aphasic
stroke. Brain 2005;128:2858-71.
51. Price CJ. The anatomy of language: a review of 100 fMRI
studies published in 2009. Ann N Y Acad Sci 2010;1191:62-88.
52. Hajnal JV, Myers R, Oatridge A, Schwieso JE, Young IR,
Bydder GM. Artifacts due to stimulus correlated motion in
functional imaging of the brain. Magn Reson Med 1994;31:
283-91.
53. Martin PI, Naeser MA, Doron KW, et al. Overt naming in
aphasia studied with a functional MRI hemodynamic delay de-
sign. Neuroimage 2005;28:194-204.
54. Cacace AT, Tasciyan T, Cousins JP. Principles of functional
magnetic resonance imaging: application to auditory neurosci-
ence. J Am Acad Audiol 2000;11:239-72.
55. Bernal B, Altman NR. Auditory functional MR imaging. AJR
Am J Roentgenol 2001;176:1009-15.
56. Johnsrude IS, Giraud AL, Frackowiak RS. Functional imaging of
the auditory system: the use of positron emission tomography.
Audiol Neurootol 2002;7:251-76.
57. Cho Z, Chung S, Lim D, Wong E. Effects of the acoustic noise
of the gradient systems on fMRI: a study on auditory, motor, and
visual cortices. Magn Reson Med 1998;39:331-5.
58. Elliott MR, Bowtell RW, Morris PG. The effect of scanner sound
in visual, motor, and auditory functional MRI. Magn Reson Med
1999;41:1230-5.
59. Lövblad KO, Thomas R, Jakob PM, et al. Silent functional
magnetic resonance imaging demonstrates focal activation in
rapid eye movement sleep. Neurology 1999;53:2193-5.
60. Amaro E Jr, Williams SC, Shergill SS, et al. Acoustic noise and
functional magnetic resonance imaging: current strategies and
future prospects. J Magn Reson Imaging 2002;16:497-510.
61. Hall D, Haggard M, Akeroyd M, et al. “Sparse” temporal sam-
pling in auditory fMRI. Hum Brain Mapp 1999;7:213-23.
62. Schmidt CF, Zaehle T, Meyer M, Geiser E, Boesiger P, Jancke
L. Silent and continuous fMRI scanning differentially modulate
activation in an auditory language comprehension task. Hum
Brain Mapp 2008;29:46-56.
63. Edmister WB, Talavage TM, Ledden PJ, Weisskoff RM. Im-
proved auditory cortex imaging using clustered volume acquisi-
tions. Hum Brain Mapp 1999;7:89-97.
64. Yetkin FZ, Roland PS, Purdy PD, Christensen WF. Evaluation of
auditory cortex activation by using silent FMRI. Am J Otolar-
yngol 2003;24:281-9.
65. Kovacs S, Peeters R, Smits M, De Ridder D, Van Hecke P,
Sunaert S. Activation of cortical and subcortical auditory
structures at3Tbymeans of a functional magnetic resonance
imaging paradigm suitable for clinical use. Invest Radiol
2006;41:87-96.
66. Smits M, Kovacs S, de Ridder D, Peeters RR, van Hecke P,
Sunaert S. Lateralization of functional magnetic resonance im-
aging (fMRI) activation in the auditory pathway of patients with
lateralized tinnitus. Neuroradiology 2007;49:669-79.
67. Bonakdarpour B, Parrish TB, Thompson CK. Hemodynamic
response function in patients with stroke-induced aphasia: impli-
cations for fMRI data analysis. Neuroimage 2007;36:322-31.
68. Thompson CK, den Ouden DB, Bonakdarpour B, Garibaldi K,
Parrish TB. Neural plasticity and treatment-induced recovery of
sentence processing in agrammatism. Neuropsychologia 2010;
48:3211-27.
69. Carusone LM, Srinivasan J, Gitelman DR, Mesulam MM, Par-
rish TB. Hemodynamic response changes in cerebrovascular
disease: implications for functional MR imaging. AJNR Am J
Neuroradiol 2002;23:1222-8.
70. Binkofski F, Seitz RJ. Modulation of the BOLD-response in
early recovery from sensorimotor stroke. Neurology 2004;63:
1223-9.
71. Liu HL, Huang JC, Wu CT, Hsu YY. Detectability of blood
oxygenation level-dependent signal changes during short breath
hold duration. Magn Reson Imaging 2002;20:643-8.
S13MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
Author's personal copy
72. Eskey CJ, Sanelli PC. Perfusion imaging of cerebrovascular
reserve. Neuroimaging Clin N Am 2005;15:367-81, xi.
73. Chen EE, Small SL. Test-retest reliability in fMRI of language:
group and task effects. Brain Lang 2007;102:176-85.
74. Andersen SM, Rapcsak SZ, Beeson PM. Cost function masking
during normalization of brains with focal lesions: still a neces-
sity? Neuroimage 2010;53:78-84.
75. Crinion J, Ashburner J, Leff A, Brett M, Price C, Friston K.
Spatial normalization of lesioned brains: performance evaluation
and impact on fMRI analyses. Neuroimage 2007;37:866-75.
76. Stephan KE, Mattout J, David O, Friston KJ. Models of func-
tional neuroimaging data. Curr Med Imaging Rev 2006;2:15-34.
77. Friston K. Causal modelling and brain connectivity in functional
magnetic resonance imaging. PLoS Biol 2009;7:e33.
78. Friston KJ, Harrison L, Penny W. Dynamic causal modelling.
Neuroimage 2003;19:1273-302.
79. Lee L, Friston K, Horwitz B. Large-scale neural models and
dynamic causal modelling. Neuroimage 2006;30:1243-54.
80. Stephan KE, Harrison LM, Kiebel SJ, David O, Penny WD,
Friston KJ. Dynamic causal models of neural system dynamics:
current state and future extensions. J Biosci 2007;32:129-44.
81. Londei A, D’ausilio A, Basso D, et al. Sensory-motor brain
network connectivity for speech comprehension. Hum Brain
Mapp 2010;31:567-80.
82. Warren JE, Crinion JT, Lambon Ralph MA, Wise RJ. Anterior
temporal lobe connectivity correlates with functional outcome
after aphasic stroke. Brain 2009;132:3428-42.
83. Detre JA, Alsop DC. Perfusion magnetic resonance imaging with
continuous arterial spin labeling: methods and clinical applica-
tions in the central nervous system. Eur J Radiol 1999;30:115-24.
84. Horwitz B, Braun AR. Brain network interactions in auditory,
visual and linguistic processing. Brain Lang 2004;89:377-84.
85. Horwitz B, Warner B, Fitzer J, Tagamets MA, Husain FT, Long
TW. Investigating the neural basis for functional and effective
connectivity. Application to fMRI. Philos Trans R Soc Lond B,
Biol Sci 2005;360:1093-108.
86. Huisman TA. Diffusion-weighted imaging: basic concepts and
application in cerebral stroke and head trauma. Eur Radiol 2003;
13:2283-97.
87. Basser PJ, Jones DK. Diffusion-tensor MRI: theory, experimen-
tal design and data analysis - a technical review. NMR Biomed
2002;15:456-67.
88. Pierpaoli C, Jezzard P, Basser PJ, Barnett A, Di Chiro G.
Diffusion tensor MR imaging of the human brain. Radiology
1996;201:637-48.
89. Mori S, van Zijl PC. Fiber tracking: principles and strategies - a
technical review. NMR Biomed 2002;15:468-80.
90. Mori S, Crain BJ, Chacko VP, van Zijl PC. Three-dimensional
tracking of axonal projections in the brain by magnetic resonance
imaging. Ann Neurol 1999;45:265-9.
91. Jones DK, Horsfield MA, Simmons A. Optimal strategies for
measuring diffusion in anisotropic systems by magnetic reso-
nance imaging. Magn Reson Med 1999;42:515-25.
92. Masutani Y, Aoki S, Abe O, Hayashi N, Otomo K. MR diffusion
tensor imaging: recent advance and new techniques for diffusion
tensor visualization. Eur J Radiol 2003;46:53-66.
93. Jones DK. Studying connections in the living human brain with
diffusion MRI. Cortex 2008;44:936-52.
94. Behrens TE, Woolrich MW, Jenkinson M, et al. Characterization
and propagation of uncertainty in diffusion-weighted MR imag-
ing. Magn Reson Med 2003;50:1077-88.
95. Catani M, Mesulam M. The arcuate fasciculus and the discon-
nection theme in language and aphasia: history and current state.
Cortex 2008;44:953-61.
96. Vernooij MW, Smits M, Wielopolski PA, Houston GC, Kres-
tin GP, van der Lugt A. Fiber density asymmetry of the
arcuate fasciculus in relation to functional hemispheric lan-
guage lateralization in both right- and left-handed healthy
subjects: a combined fMRI and DTI study. Neuroimage 2007;
35:1064-76.
97. Parker GJ, Luzzi S, Alexander DC, Wheeler-Kingshott CA,
Ciccarelli O, Lambon Ralph MA. Lateralization of ventral and
dorsal auditory-language pathways in the human brain. Neuro-
image 2005;24:656-66.
98. Saur D, Kreher BW, Schnell S, et al. Ventral and dorsal path-
ways for language. Proc Natl Acad Sci USA 2008;105:18035-40.
99. Catani M, Jones DK, ffytche DH. Perisylvian language networks
of the human brain. Ann Neurol 2005;57:8-16.
100. Jones DK, Williams SC, Gasston D, Horsfield MA, Simmons A,
Howard R. Isotropic resolution diffusion tensor imaging with
whole brain acquisition in a clinically acceptable time. Hum
Brain Mapp 2002;15:216-30.
S14 MAGNETIC RESONANCE IMAGING OF LANGUAGE RECOVERY, Smits
Arch Phys Med Rehabil Vol 93, Suppl 1, January 2012
    • "Functional MRI (fMRI) has been increasingly used to assess the functional alterations in stroke. fMRI is a powerful and non-invasive tool to delineate functional abnormality in the brain after stroke1234 . Earlier studies have mainly used task-based fMRI and identified abnormal local activities and inter-regional connectivity involving the fronto-parietal motor control sys- tems [5], the sensorimotor cortex67, inferior parietal lobule region and cerebellum regions [8], temporo-parietal area [7], fronto-temporo-occipital area, supramarginal gyrus, angular gyrus, and basal ganglia and fusiform gyrus [9]. "
    [Show abstract] [Hide abstract] ABSTRACT: Resting-state functional magnetic resonance imaging (R-fMRI) has been intensively used to assess alterations of inter-regional functional connectivity in patients with stroke, but the regional properties of brain activity in stroke have not yet been fully investigated. Additionally, no study has examined a frequency effect on such regional properties in stroke patients, although this effect has been shown to play important roles in both normal brain functioning and functional abnormalities. Here we utilized R-fMRI to measure the amplitude of low-frequency fluctuations (ALFF) and regional homogeneity (ReHo), two major methods for characterizing the regional properties of R-fMRI, in three different frequency bands (slow-5: 0.01-0.027 Hz; slow-4: 0.027-0.73 Hz; and typical band: 0.01-0.1 Hz) in 19 stroke patients and 15 healthy controls. Both the ALFF and ReHo analyses revealed changes in brain activity in a number of brain regions, particularly the parietal cortex, in stroke patients compared with healthy controls. Remarkably, the regions with changed activity as detected by the slow-5 band data were more extensive, and this finding was true for both the ALFF and ReHo analyses. These results not only confirm previous studies showing abnormality in the parietal cortex in patients with stroke, but also suggest that R-fMRI studies of stroke should take frequency effects into account when measuring intrinsic brain activity.
    Full-text · Article · Apr 2015
    • "The arcuate fasciculus connects the Broca's area (a frontal expressive language area) with the Wernicke's area (a posterior temporoparietal deceptive language area), and is the best known language related white matter pathway (Smits et al., 2012). Although we did not analyze the specific white matter involvement, our observation of white matter involvement is in agreement with those of previous studies (Naeser and Palumbo, 1994, Okuda et al., 1994, Smits et al., 2012). On the contrary, Bates et al. (2003) revealed lesions in the insula and arcuate/superior longitudinal fasciculus most affect speech production, and lesions in the middle temporal gyrus most affect speech comprehension, neither of which appeared in our results (Table 1). "
    [Show abstract] [Hide abstract] ABSTRACT: Global aphasia without hemiparesis is a striking stroke syndrome involving language impairment without the typically manifested contralateral hemiparesis, which is usually seen in patients with global aphasia following large left perisylvian lesions. The objective of this study is to elucidate the specific areas for lesion localization of global aphasia without hemiparesis by retrospectively studying the brain magnetic resonance images of six patients with global aphasia without hemiparesis to define global aphasia without hemiparesis-related stroke lesions before overlapping the images to visualize the most overlapped area. Talairach coordinates for the most overlapped areas were converted to corresponding anatomical regions. Lesions where the images of more than three patients overlapped were considered significant. The overlapped global aphasia without hemiparesis related stroke lesions of six patients revealed that the significantly involved anatomical lesions were as follows: frontal lobe, sub-gyral, sub-lobar, extra-nuclear, corpus callosum, and inferior frontal gyrus, while caudate, claustrum, middle frontal gyrus, limbic lobe, temporal lobe, superior temporal gyrus, uncus, anterior cingulate, parahippocampal, amygdala, and subcallosal gyrus were seen less significantly involved. This study is the first to demonstrate the heterogeneous anatomical involvement in global aphasia without hemiparesis by overlapping of the brain magnetic resonance images.
    Full-text · Article · Dec 2014
    • "A shift from a model-led to a phenotype-led approach may clarify whether any classification scheme is indeed clinically meaningful. Also, aphasia studies using more advanced magnetic resonance neuroimaging techniques and functional imaging tools such as functional MRI, PET, SPECT, and DTI may further elucidate the relationship between lesion site and language impairment in the context of such an approach.4,78,85 "
    [Show abstract] [Hide abstract] ABSTRACT: One of the most devastating consequences of stroke is aphasia. Communication problems after stroke can severely impair the patient's quality of life and make even simple everyday tasks challenging. Despite intense research in the field of aphasiology, the type of language impairment has not yet been localized and correlated with brain damage, making it difficult to predict the language outcome for stroke patients with aphasia. Our primary objective is to present the available evidence that highlights the difficulties of predicting language impairment after stroke. The different levels of complexity involved in predicting the lesion site from language impairment and ultimately predicting the long-term outcome in stroke patients with aphasia were explored. Future directions and potential implications for research and clinical practice are highlighted.
    Full-text · Article · Apr 2014
Show more