Diffusion Tensor Imaging

Johansen-Berg and Rushworth 2008Glasser and Rilling 2008

20 April 2010

Connectivity

• Connectivity patterns define functional networks– Passingham et al. (2002)

• Connectional anatomy– Boiling in oil– Dissection– Myelin stains– Axonal stains– Retrograde and anterograde tracers

Diffusion-weighted MRI

• Acquisition of multiple images• Each image is sensitized to a different direction

so multiple measures are associated with each voxel

• Fit a mathematical model such as diffusion tensor imaging– Each voxel is described by an ellipsoid, or tensor– Three orthogonal eigenvectors and associated

eigenvalues (λ1, λ2, λ3)

Diffusion-weighted MRI

• Self-diffusion of water molecules• Water is directionally dependent in tissue

with directional structure

Fractional Anisotropy

• In the ventricles, diffusion should be similar in each direction which is associated with a spherical tensor

• Fiber bundles have more barriers and the associated tensor is elongated

• Directional dependency (Anisotropy) is quantified by fractional anisotropy

• 0 (fully isotropic) to 1 (fully anisotropic)– White matter: ~0.6-0.8

Diffusion Anisotropy

• Directional estimates of principal diffusion direction form the basis of diffusion tractography

• This is NOT anatomical• Functional connectivity is just the correlation

between two or more active areas

Deterministic vs. Probabilistic Models

• Difficult to interpolate beyond the end of the fiber bundle

• One can assign a probability to the direction of each voxel

• Sampling is used to create streamlines which are used to represent interconnectivity

Strengths and Weaknesses

• Trace cortico-cortical networks– Parietal and premotor

cortex• Fine-grained spatial

mapping– Thalamic connections

• Difficult to determine connection strength

• Anterograde or retrograde?

• Accurate detection of connections

• Validation: biologically realistic phantoms

DTI Tractography of the Human Brain’s Language

Pathways

Glasser and Rilling 2008http://cercor.oxfordjournals.org/

cgi/content/full/bhn011/DC1

DTI and Language

• Arcuate Fasciculus– Originates in temporal lobe– Curves around Sylvian fissure– Projects into the frontal lobe– Connects Broca’s area and Wernicke’s area– Dejerine 1895

• Leftward asymmetry in the arcuate fasciculus

Three Components of Langauge

• Phonemes– Wernicke’s area and

BA 40– Posterior Broca’s area– Bilateral

• Prosody– Right MTG– Frontal lobe– Bilateral?

• Lexical-semantics– Concepts and

meanings– Middle and temporal

areas (BA 21 and 17)– Broca’s area and

areas more anterior and frontal

– Left localized

Hypotheses

• Bilateral arcuate connection to the superior temporal gyrus

• Leftward asymmetry connection to middle temporal gyrus

Methodology

• No behavioral tasks• A priori reasons to think phonology, lexico-

semantics, and prosody have distinct locations

• Determine the connections using a deterministic algorithm

• Corroborate the connections using outside sources

Subjects

• 20 right-handed males• 18-50 (23.75, 7.1)• Handedness and sex influence laterality

MRI

• Diffusion MRI• TE = 90 ms• TR = 7700 ms• 12 diffusion directions• 1.7 x 1.7 x 2.0 mm

voxels

• T1-weighted• TE = 4 ms• TR = 2300 ms• 256 x 256 matrix

Deterministic Tractography

• Siemens DTI Task Card Parameters• # of samples/voxel length = 8• Minimum FA threshold = 0.15• Maximum turning angle = 15°• Step length between calculations = 0.25

mm

Deterministic Tractography

• Step 1:– Select Arcuate Fasciculus as ROI in left and

right hemispheres from a coronal slice• ROIs defined as BA 22 (Wernicke’s), BA

21 and BA 37 (middle temporal gyrus)• Determine terminations in frontal lobe ROI• BA 44, 45, 6, 9 (Broca’s area and

surrounding cortex)

ROIs

Arcuate Fasciculus Tractography

The STG Pathway

• Left Hemisphere• Found in 17 of 20

subjects• BA 22 (posterior

STG) connected to BA 44 and BA 6

• Right Hemisphere• Found in 4 of 20

subjects• BA 22 (posterior

STG) connected to BA 44 and BA 6

The MTG Pathway

• Left Hemisphere• Found in 20 of 20• BA 21 and BA 37

projected to• BA 44• and parts of BA 6, 9,

45

• Right Hemisphere• Found in 11 of 20• BA 37 (posterior

MTG)

Functional Activations

• Left Hemisphere• Phonological Studies

overlapped with STG pathway

• Lexical-Semantic Studies overlapped with MTG, ITG, and AG segments

• Right Hemisphere• Phonological Studies

did not overlap (more anterior) with STG pathway

• Prosody Studies overlapped MTG and STG segments

Asymmetries in Connections

• Phonology is usually bilaterally activated; more activation found on the left side than usual

• Lexical-Semantics is usually activated on the left; more activation found on the right than usual

Hickok and Poeppel (2004) and Price (2000) Model: Left

Hickok and Poeppel (2004) and Price (2000) Model: Right

Aphasia

• Can we explain aphasia using the Hickok and Poeppel (2004) and Price (2000) model with additional information added from the DTI study?

Broca’s Aphasia

• Symptoms– Difficulty producing grammatical speech– Slow and halting speech– Able to communicate– Difficulty with phoneme discrimination

• Location of lesions– Left inferior and premotor cortex (Dysarthria)– BA 44 and BA 45 (agrammatism)– Frontal cortical terminations

Wernicke’s Aphasia

• Symptoms– Fluent speech– Frequent phonological and semantic errors– Difficulty understanding speech

• Location of lesions– Left STG or subcortical connections– Middle and inferior temporal cortex

• Disruption of phonological decoding

Conduction Aphasia

• Symptoms– Difficulty with repetition– Phonological errors– Difficulty naming objects– Intact comprehension– Spontaneous speech

• Location of lesions– Left supramarginal gyrus (BA 40)– Arcuate fasciculus?

• Superficial to avoid damaging the MTG pathway

Conduction Aphasia

• Symptoms– Difficulty with repetition– Difficulty naming objects– Phonological errors– Intact comprehension– Spontaneous speech preserved

• Location of lesions– Left supramarginal gyrus (BA 40)– Superficial arcuate fasciculus

Transcortical Motor Aphasia

• Symptoms– Limited spontaneous speech– Impaired naming– Intact repetition– Normal articulation– Good auditory comprehension

• Location of lesions– Anterior and superior to Broca’s area– White matter near MTG pathway

Transcortical Sensory Aphasia

• Symptoms– Intact repetition– Lack of understanding

• Location of lesions– Left temporal lobe lesion which spares

Wernicke’s area– Lesions in middle and inferior temporal lobe

• Electrical stimulation of LMTG and LITG causes transient aphasia

Aprosodias

• Symptoms– Difficulty with comprehension of emotional

prosody• Location of lesions

– Right temperoparietal lesions– Right middle and temporal cortex– Subcortical

Limitations

• Unidentified pathways probably exist• Crossing fibers add noise• Deterministic vs. Probabilistic• Heterogeneity of functional areas• Only three components of language

Unexpectations

• Lexical-semantic and STG pathways project to BA 6

• Fewer projections to BA 45 and BA 47 using a deterministic methodology than a probabilistic methodology

• More anterior with probabilistic• These differences could be explained if the

arcuate fasciculus was combined with the SLF

Conclusions

• Arcuate fasciculus connects the STG and MTG with the inferior frontal lobe

• Phonological processing associated with left STG

• Lexical-semantic processing associated with left MTG

• Prosody processing associated with right MTG and STG

Hickok and Poeppel (2004) and Price (2000) Model: Aphasia

Questions

• Marianna– How can they find evidence of activation of

different areas?• Lucy

– What would these models predict for people who are relatively more bilateral? Just more/higher incidence of pathways in the right hemisphere?

Questions

• Israel– How do they know that the connections they find

are related to these linguistic areas and not to the other linguistic areas?

• Pawel – Couldn't familial left-handedness explain the

results showing that they were unable to find MTG pathways in the right hemisphere of some subjects and the STG pathway in the left hemisphere of others?

Questions

• Lynn– Could the motor theory or elements of it be

useful in interpreting any of the current results including visual information and the right hemisphere?

Neat Website

• http://white.stanford.edu/~brian/papers/mri/2006-Wandell-NIPS-Tutorial.pdf