Chapter 2: Experiment 1: Callosal tracts and patterns of hemispheric dominance1
Left-hemispheric dominance for language and right-hemispheric dominance for spatial processing are distinctive characteristics of the human brain. However, variations of these hemispheric asymmetries have been observed, with a minority showing crowding of both functions to the same hemisphere or even a mirror reversal of the typical lateralization pattern. Here, I used diffusion tensor imaging and functional magnetic imaging to investigate the role of the corpus callosum in participants with atypical hemispheric dominance. The corpus callosum was segmented according to the projection site of the underlying fibre tracts. Analyses of the microstructure of the identified callosal segments revealed that atypical hemispheric dominance for language was associated with high anisotropic diffusion through the corpus callosum as a whole. This effect was most evident in participants with crowding of both functions to the right. The enhanced anisotropic diffusion in atypical hemispheric dominance implies that in these individuals the two hemispheres are more heavily interconnected.
The corpus callosum is the largest fibre tract in the human brain connecting the left and right hemispheres and relaying sensory, motor, and higher cognitive information. It is commonly suggested that the corpus callosum plays an important role in the development and maintenance of functional hemispheric asymmetries. The best documented functional asymmetry is the left-hemispheric dominance for language, first described by Paul Broca in the 1860s (as cited in Berker et al., 1986). Non-verbal functions such as spatial attention, face processing, mental rotation, and emotion processing are considered to be right-hemisphere dominant (Bourne, 2008; Corballis, 1997; Vogel, Bowers, & Vogel, 2003).
Handedness is positively correlated with hemispheric dominance for language production, but not with lateralization for spatial processing (Badzakova-Trajkov, Haberling, Roberts, et al., 2010). The correlation with language dominance has led some to propose a common genetic origin. Genetic models, such as those proposed by Annett (1998, 2002) and McManus and Bryden (1992), assume a single gene with one allele predisposing to right-handedness and left-hemispheric dominance for language, and the other leaving the direction of both asymmetries to chance.
Two contrasting models have been suggested for the role of the corpus callosum in brain asymmetry. According to the excitatory model, the corpus callosum integrates information between the hemispheres (Galaburda et al., 1990; Gazzaniga, 2000), so that bilateral processing is associated with enhanced interhemispheric transfer in order to maintain both hemispheres in a functionally activated state. In contrast, the inhibitory model proposes that fibres in the corpus callosum inhibit homotopic areas, allowing for independent functioning of the hemispheres (Cook, 1984a). Thus, where the excitatory model predicts a negative correlation between degree of lateralization and callosal connectivity, the inhibitory model predicts a positive correlation.
Studies based on handedness have been equivocal, with some showing larger callosal areas in left- than in right-handers (Habib et al., 1991; Witelson, 1985, 1989; Witelson & Goldsmith, 1991), while others report no differences (Clarke & Zaidel, 1994; Jancke, Staiger, et al., 1997; Preuss et al., 2002; Steinmetz et al., 1992), or even larger callosal areas in right-handers (Westerhausen et al., 2004). Where language lateralization was measured more directly, for example with the dichotic listening task, correlations between degree of lateralization and the midsagittal area of the corpus callosum have been negative (Gootjes et al., 2006; Hines et al., 1992; Yazgan et al., 1995), favouring the excitatory model. Similarly, bilateral processing of spatial stimuli, as indicated by a smaller leftward bias in the line-bisection test, was associated with larger callosal areas (Yazgan et al., 1995).
Using fMRI during word generation to measure language lateralization and DTI to examine the microstructure of the corpus callosum, Westerhausen et al. (2006) found that mean diffusivity (MD) in the corpus callosum was lower, while fractional anisotropy (FA) tended to be higher, in strongly left-lateralized subjects than in moderately left-lateralized, 36 bilateral, or right-lateralized subjects, but the groups did not differ with respect to the area of the corpus callosum. The authors interpreted the MD and FA results as indicating stronger and / or faster interhemispheric connections in the strongly left-lateralized individuals, which might be taken as support for the inhibitory model. Josse et al. (2008) found that increased callosal size was associated with greater left-hemisphere activity during language tasks in posterior temporal and inferior frontal regions, together with reduced right-hemisphere activity in posterior temporal regions, again supporting an inhibitory model. In this study, though, participants with right-hemispheric language dominance were excluded.
From an evolutionary perspective, Gazzaniga (2000) has proposed that callosal function was reduced in humans relative to that in primates, to allow differential specialization of the hemispheres. Nevertheless, sufficient interhemispheric connections were retained to permit computations in the two hemispheres to be integrated. On this view, one would predict reduced callosal function in people with the typical pattern of cerebral asymmetry, including left-hemisphere dominance for language and right-hemisphere dominance for spatial processing, compared to those with atypical cerebral asymmetry, especially where the same hemisphere is dominant for both language and spatial processing. Here, I aim to test this prediction using fMRI to assess lateralization for language and spatial processing, and fractional anisotropy (FA) to measure the efficacy of callosal projections. Increased FA can be found in white matter with more tightly packed axons, thicker myelin sheath, fewer obliquely oriented fibres, and different radii of individual axons (Chepuri et al., 2002). My main hypothesis is that FA should be higher in those with atypical patterns of cerebral asymmetry.
Materials and methods
A total of 60 participants took part in the study (30 males; mean age = 20.83, SD = 4.6; 30 females; mean age = 23.23, SD = 4.55). To establish handedness, all participants filled out a handedness questionnaire (Annett, 2002), consisting of 12 different questions on hand preference (writing, throwing a ball, holding a racquet, lighting a match, cutting with scissors, threading through a needle, broom sweeping, shovelling, dealing cards, hammering, holding a toothbrush, unscrewing a lid) (Appendix A). They were asked to indicate the hand habitually used for each of those activities by giving two ticks for a preferred hand or one tick for each hand when indifferent. Preference scores were obtained from the formula 100 ( + − ), where R and L are the numbers of ticks allocated to the right and left hands, respectively. 35 participants with preference scores > 0 (17 male) were classified as right-handers whereas 25 (13 males) with preference scores ≤ 0 were classified as left-handers2. Left-handers were deliberately over-represented in the study to allow for comparisons between handedness groups and increase the chance of including individuals with atypical patterns of hemispheric dominance.
Most participants were undergraduate students at the University of Auckland and had no history of psychiatric or neurological disorders (Appendix B). They all gave written consent prior to the study and ethical approval was obtained by the Human Ethics Participants Committee at the University of Auckland.
Word Generation Task
Experiments were programmed using E-Prime software (http://www.pstnet.com/). For the Word Generation Task, participants were instructed to silently generate as many words as possible beginning with a designated letter (F, A, S, B, and M) projected in random order and centrally onto a screen (in Courier New font, size 50). Participants were advised to avoid proper names or the same words with different endings. Experimental blocks were followed by a baseline period that consisted of a small black fixation cross. Baseline and experimental blocks lasted 30 s and were repeated 5 times resulting in a scanning time of 5 min. To obtain a behavioural performance measure, each participant undertook a comparable verbal fluency test with 3 letters (P, R, and W) outside the scanner, in which they spoke the words out loud.
The Landmark Task is a variant of a line-bisection task, which has been shown to engage a predominantly right-hemispheric network in parietal and occipital cortices (Fink et al., 2000). It consisted of an experimental condition during which participants decided whether a horizontal line was bisected exactly in the middle, and a control condition where participants had to judge whether a line was bisected at all. The stimuli were lines of three different lengths, 5 cm, 8 cm and 10 cm, resulting in a visual angle of 11, 22 and 33°, respectively. Stimuli were presented centrally on a white screen for 1s with an inter-stimulus interval of 1s during which time participants responded to the task by pressing one of two buttons with either their left or their right hand. They pressed with the index finger for answering “yes” to either the question “Is the line bisected exactly in the middle?” or “Is there a vertical mark?”, and with the middle finger for answering “no.” The lines were correctly bisected in the middle on 50% of trials, and deviated in 25% of the trials to the left and in 25% of the trials to the right with a bias of 2, 5 or 10% of the lengths of the line. In the control condition, the line was bisected in half of the trials, while in the other half no vertical mark was presented. Each experimental and control block lasted 30 s and was followed by a baseline of 12.5 s that consisted of a black fixation cross.
An introduction screen presented for 5 s gave instructions as to which hand to use and which condition should be performed. The ensuing experimental or control block was repeated 3 times before a new instruction screen appeared. This scheme was repeated 4 times (2 conditions x 2 hands) so that all possible combinations of hand and condition were performed three times by all subjects. The order was counterbalanced. Total scanning time was 8 min 50 s. Prior to the scanning, participants were given a practice version of the task with 2 bisection and 2 control blocks in order to familiarize them with the task.
All MRI scanning was performed on a 1.5-T Siemens Avanto scanner (Erlangen, Germany). A T1-image was acquired using a 3-D MP-RAGE sequence with the following parameters: TR = 11 msec; TE = 4.94 msec; flip angle: 158; FOV: 256 x 256 mm2; up to 176 axial slices, ensuring whole brain coverage, parallel to AC–PC line; slice thickness: 1 mm; interslice gap: 0 mm resulting in 1 x 1 x 1 mm voxels. For the functional images an EPI sequence with the following parameters was used: TR = 2500 ms; TE = 50 ms; flip angle = 90; FOV = 192 x 192 mm2; matrix size: 64 x 64; 29 slices parallel to the AC–PC line; slice thickness: 3 mm; and interslice gap: 25% = 0.8 mm. Diffusion-weighted data were acquired using a single-shot spin echo sequence (TR = 6601 ms, TE = 101 ms, in-plane resolution: 1.8 x 1.8 mm, and FOV 230 mm). 45 axial slices (thickness 3 mm) along 30 diffusion gradient directions (b = 1000 s/mm2) as well as one image without diffusion weighting (b = 0 s/mm2) were acquired. This sequence was repeated twice resulting in an acquisition time of approximately 7 min.
Image pre-processing and analyses
Functional data were analysed using SPM5 software (Wellcome Trust Centre for Neuroimaging, London, UK; www.fil.ion.ucl.ac.uk/spm). The first two scans were discarded to allow for saturation of the MRI signal. The functional scans were realigned to the first image of the session and the T1-weighted structural image was co-registered to the mean of the functional volumes. Both, structural and functional images were normalized into standardised stereotactic space (MNI, Montreal Neurological Institute) and spatially smoothed with an anisotropic Gaussian filter of 9 x 9 x 9 mm of full-width at half maximum (FWHM). For each participant a first-level analysis was performed using the general linear model (GLM) implemented in the SPM5 software. The models were set up as a box-car function with “task” vs. “control” conditions (letter vs. baseline for the Word Generation Task; bisection vs. control for the Landmark Task) and convolved with a canonical haemodynamic response function. Movement regressors were also included in the model. To see the general pattern of activation, a second-level random-effects analysis was performed by applying a one-sample t-test to the contrast images of the first-level analyses. To correct for multiple comparisons, a family-wise error (FWE) correction was applied at p < 0.05 with a contiguity threshold of 20 voxels.
For each participant, laterality indices were calculated using the laterality toolbox available on the SPM website (Wilke & Lidzba, 2007). Laterality was scored after the widely-used formula LI = � + − �, where L and R represent activation in left and right hemisphere, respectively. Because laterality indices based on one single fixed statistical threshold do not yield robust or reproducible results (Jansen et al., 2006), the LI toolbox applies a bootstrap algorithm to calculate about 10,000 indices at different thresholds yielding a robust mean, and maximum and minimum LIs. Indices range from -1 for extreme right to 1 for extreme left lateralization. Regions of interest (ROI) were defined anatomically within the toolbox. In addition, the toolbox applies a weighting factor to take asymmetric mask sizes into account. For the Word Generation Task the whole frontal lobe and for the Landmark Task the whole parietal lobe were considered. The laterality indices were used to decide whether the language processing was left dominant, symmetric or right dominant (Fernandez et al., 2003). A positive index within the ROI corresponded to a left-hemispheric dominance (LI > 0.1) and a negative index to a right-hemispheric dominance (LI < -0.1). Bilateral processing was characterized by LIs ranging from -0.1 to +0.1.
Although there has been some question as to the reliability of fMRI-based laterality measures (Jansen et al., 2006), classification of language lateralization based on this task has yielded results consistent with the Wada test, which is thought to be the “gold standard” for measuring hemispheric dominance (Woermann et al., 2003; Yetkin et al., 1998), and also yield results consistent with fTCD (Jansen et al., 2004). Moreover, the test-retest reliability of lateralization has been shown to be high (Adcock, Wise, Oxbury, Oxbury, & Matthews, 2003; Fernandez et al., 2003).
Diffusion-weighted image analysis and tractography based segmentation
DTI was used to divide the corpus callosum based on the projection site of specific fibre tracts (Chao et al., 2009; Zarei et al., 2006). The tractography-based segmentation of the corpus callosum was carried out using the FDT toolbox implemented in FSL (http://www.fmrib.ox.ac.uk/fsl/). Based on the probability that a voxel in a seed mask connects to a predefined target mask, the anatomical connectivity pattern of a brain region can be determined (Behrens et al., 2003). This method was applied to the callosal fibre tracts as follows.
For each participant, the corpus callosum was manually outlined on the midsagittal slice of the individual T1 images by applying the following steps. First, the structural T1-weighted images were registered to an optimized individual target brain in the Talairach coordinate system (Kochunov et al., 2002) available at the BrainMap website (http://brainmap.org/ale/index.html). A rigid body transformation with 6 degrees of freedom (3 translations, 3 rotations) was applied without scaling the brain in order to maintain individual brain sizes. Second, the images were segmented into grey and white matter using the automated segmentation tool implemented in FSL (Zhang, Brady, & Smith, 2001). Third, the white matter segmentation image was overlaid over the structural image to facilitate the manual outlining of the corpus callosum. To assess reliability, a second independent measurement of the callosal area for 10 randomly chosen participants was made by a different rater, resulting in an intraclass correlation of 0.96. In addition, the white matter volume was calculated in order to be able to control callosal measurements for brain size.
To segment the corpus callosum into three distinct subdivisions, specific cortical target masks were created on the MNI template. The most anterior mask comprised the frontal lobe but spared the motor cortex. The motor cortex was derived from the Juelich histological atlas implemented in FSL and included M1 and premotor cortex. The posterior mask included the temporal, parietal and occipital lobes. To ensure a good alignment of these masks, the individual T1 images were nonlinearly registered to the MNI template and the resulting registration matrices were inversed and applied to the target masks transferring them back into each participant’s native space.
The diffusion data were corrected for eddy currents and head motion and the two acquisitions were averaged to improve signal-to-noise ratio. Diffusion tensors were fitted at each voxel and FA maps were generated. Then, the probability distribution of fibre direction at each voxel was calculated using previously described methods (Behrens et al., 2003). For each participant, probabilistic tractography was run from each voxel in the corpus callosum to each target mask by drawing 5000 random samples with a step length of 0.5 mm and a curvature threshold of 0.2. The resulting value in each callosal voxel represents the number of samples reaching the relevant target mask and therefore the probability of the connections between the two. The analysis was run separately for the right and left hemispheres. Using a similar threshold as in previous research, each output mask was thresholded at 10% of the maximum connectivity value to the relevant target mask (Zarei et al., 2006). Then, tracts going to the left and right hemisphere were compared and only overlapping areas were considered for further analyses (Westerhausen et al., 2009). The resulting masks were binarized and multiplied with the FA maps in order to extract the mean FA values of the corpus callosum.
To visualize the results, the callosal segments were registered to the MNI template by applying the nonlinear registration matrices calculated before, and then added together. This generates a population-based probability mask where the values in each voxel represent the number of participants that show a connection to the relevant target mask. A threshold was applied so that the resulting masks only contain voxels that were shared by at least 70% of the participants. This threshold was arbitrary and only used for the visualization of the resulting callosal segments. The tractography based segmentation of the corpus callosum is visualized in Figure 5.
Chapter 1: General introduction
Origins of hemispheric specialization
The Geschwind-Galaburda model of hemispheric specialization
Structural connectivity in the asymmetric brain
Chapter 2: Experiment 1: Callosal tracts and patterns of hemispheric dominance
Materials and methods
Chapter 3: Experiment 2: The corpus callosum in monozygotic twins concordant and discordant for handedness and language dominance
Material and Methods
Chapter 4: Experiment 3: Asymmetries of the arcuate fasciculus in monozygotic twins: Genetic and nongenetic influences
Material and methods.
Chapter 5: General discussion .
The relationship between structural connectivity and hemispheric specialization
Genetic and nongenetic factors in structural connectivity
Gender and handedness effects
Limitations, implications and future research
Summary and conclusion
GET THE COMPLETE PROJECT