Bachelor thesis
Functional Connectivity in the Resting Brain and its Relationship to Brain Structure
Name: Sabrina Okx
Student Number: 10003422 Date: 09-05-2014
Supervisor: Olympia Colizoli Number of Words: 5559
Contents
Abstract 3
Functional and Structural Connectivity 4 Resting State Functional Connectivity and Structural Connectivity Relationships 5 Used Methods and Resting State Functional Connectivity and Brain Structure Relationships 8 Variation in Functional and Structural Connectivity Relationships in Group Comparisons 11
Conclusion and Discussion 15
Literature 17
Abstract
Since recent technical developments made it relatively simple to examine the human brain in vivo, resting state Functional Connectivity (rsFC) and its relationship to Structural Connectivity (SC) has been studied widely. To assess what these studies have found so far, this thesis
investigates to what extent it is possible to predict functional connectivity from structural connectivity in the resting human brain. Although it was hypothesized that rsFC would reflect SC, this seemed to be only partly true. There is overlap in rsFC and underlying SC and it is even possible to predict rsFC from SC, but the reverse appears not to be possible. Especially when comparing different groups (e.g. sex, age, patients), rsFC and SC relationships sometimes decrease or even vanish. These findings have important implications for future research within this field of research.
Functional and Structural Connectivity
Since the emergence of functional Magnetic Imaging (fMRI) studies in the 90s of the last century, it is possible to measure brain activity with a high spatial resolution. fMRI is a
neuroimaging technique based on Magnetic Resonance Imaging (MRI), where oxygenated blood flow is measured. Since oxygenated blood flow in the brain is related to neuronal activation, measuring this blood flow makes it possible to obtain completely non-invasive maps of human brain activity. Initially this technique has been widely used to investigate brain function linked to specific tasks, to have a closer look at brain areas linked to these specific tasks. The latest trend however, is to use fMRI to investigate the brain in resting state. In a resting state fMRI study, researchers are not interested in specific brain areas linked to specific tasks. On the contrary, they want to see how the brain functions when participants are instructed to just close their eyes and relax. In contrast to fMRI studies which focus on the role and function of specific brain regions, resting state fMRI studies examine functional interactions between brain regions by correlating the measured spontaneous brain activity. This resting state Functional Connectivity (rsFC) is suggested to describe the relationship between the neuronal activation patterns of anatomically separated brain regions, reflecting the level of functional communication between regions (van den Heuvel and Hulshoff Pol, 2010). Brain regions that show synchronized activity are said to form resting-state networks (RSNs) (van den Heuvel et al. 2009). The high level of resting state Functional Connectivity (rsFC) within these RSNs suggests that the involved brain areas are not only functional linked, but that there is an underlying structural pathway that provides this observed activity.
Until recently, structural connections in the brain could be examined more easily in animal studies than in the human brain. These investigations, mostly of macaque brains with the use of tract-tracer methods, have shown that structural connections between brain areas indeed seem to exist. One of the functional networks that has been investigated a lot in RSN studies is the Default Mode Network (DMN). This network covers a number of frontal and parietal areas of the brain that shows high rsFC and typically deactivates during performance of cognitive tasks (Raichle et al. 2001). Tracts connecting several nodes in the DMN have been demonstrated in Macaque studies of Structural Connectivity (SC)(Suzuki and Amaral 2003).
Recent progress in diffusion tensor imaging (DTI) made non-invasive in vivo
measurement of white matter pathways possible in human brains. DTI is an MRI technique that measures the diffusion of water molecules in the brain (Le Bihan 2003). In the brain, water diffuses more promptly along axons than across them, and measuring the direction of this diffusion can therefore be used to infer the orientation of white matter tracts in the brain. The availability of imaging techniques such as fMRI and DTI, now makes it possible to measure rsFC and SC in the human brain in a non-invasive way. As a result, nowadays it is relatively easy to investigate whether there exists a relationship between rsFC and SC in humans. This progress is reflected in the literature, the last 10 years rsFC and SC relationships have been regularly examined.
To explore what these studies have found so far, this thesis investigates to what extent it is possible to predict functional connectivity from structural connectivity in the resting human brain. Most of the studies discussed here focus on the DMN, but sometimes other RSNs will also be examined. Because monkey studies have shown a rsFC and SC relationship in the DMN before, it is hypothesized that these relationship will also exist in the human brain.
To investigate this hypothesis, the first section of this thesis will inquire whether there is a direct relationship between rsFC and SC in the human brain. The second section focuses on different methods and whether findings differ between these methods. The third section explores how findings differ between different groups of subjects.
Resting State Functional Connectivity and Structural Connectivity Relationships More than half a century ago, long before we had access to most of the current techniques in neuroscience, Hebb came up with his well-known theory. Summarized he stated that “when neurons fire together, they wire together”, which means that cells that have a synchronized electrical activity will strengthen their synaptic connections (Hebb 1949, cited in Koch et al. 2002). The Hebbian theory suggests that correlated activation patterns found in resting state fMRI studies should probably reflect underlying SC. Tract-tracer studies confirmed this inference to be at least partly true in the DMN in macaques (Suzuki and Amaral, 2003). Until recently the question remained whether this rsFC and SC relationship could also be found in the living human brain. These days the availability of DTI and fMRI techniques make it possible to set up a study where rsFC and SC relationships can be relatively easy investigated in humans. Because this opportunity is now frequently used by researchers this section explores whether there is a direct
relationship between rsFC and SC in the human brain. First an exploration throughout rsFC and SC relationships in the entire brain will be discussed, after which the focus directing will be directed to several known RSNs.
The first time rsFC and CS in the human brain where shown to be directly linked to each other was in 2002. This first examination of structural and anatomical relationships in a
quantitative way was performed by Koch, Norrit and Hund-Georgiadis (2002). In their
experiment they used 7 healthy volunteers who underwent Voxel Based Morphometry (VBM), fMRI and DTI measures in resting state. They used this measures to correlate rsFC (fMRI measures) to grey matter pixels (VBM measures) as well as to white matter tracts (DTI
measures), and compared these correlations. When comparing SC and rsFC between regions of interest on adjacent cortical gyri there seemed to be no simple correlation of these measures. Low values of rsFC were not found together with high values of SC, but high rsFC and low SC did occur together. These results indicate that regions which are clearly directly linked by white matter fiber tracts show high FC but that high FC doesn’t mean necessarily a high degree of underlying SC. Although this study indicates that there is a link between rsFC and SC in the human brain, the results are not convincing yet. The explorative approach of looking at
correlations between adjacent gyri leaves only a glimpse of what you may examine. Looking at more specific FC patterns like RSN can provide more information about rsFC and SC
relationships. Since tract-tracer studies have already demonstrated that there is a clear rsFC and SC relationship in the DMN in other primates (Suzuki and Amaral 2003), it is worth searching for this relationship in the DMN in humans too.
Likewise in human brains rsFC in the DMN appears to reflect underlying SC (Greicius, Supekar, Menon and Dougherty 2009). To test whether this hypothesis is true, Greicius et al. (2009) took resting state DTI fiber tractography and fMRI measures of 23 subjects and combined these measures in their analyses to investigate connectivity in the DMN. The DTI analyses uncovered robust SC between several parts of the DMN (including medial temporal lobes, retroplenial cortex, medial prefrontal cortex and the posterior cingulate cortex). These results indicate that rsFC reflects underlying SC to a large degree. A notable choice in this study was that only 6 out of 23 participants underwent both the fMRI and DTI measures, the other participants attended only one of the two measurements. This partly within/between subject construction points out that rsFC and SC relationships seem to be robust enough to also be found
in between subject studies. Even when performing a study in which fMRI and DTI measures where completely between subjects a simple rsFC and SC relationship is still found (Zhang, Snyder, Shimony, Fox and Raichle 2010). Zhang et al. (2010) focused in their study on how well structural information derived from diffusion-weighed MRI (a DTI measure) in a group of 12 subjects, compares with functional information derived from fMRI in a group of 17 other subjects of intrinsic brain activity in the thalamocortical system. They founded a significant overlap in connectivity profiles when compared fMRI and DTI results, percentages fluctuated around 50%. Although this shows that structural and functional mapping, performed independently (e.g. between subjects) of each other, demonstrate some overall correspondence, there is still a lot of variance in overlapping percentages. This indicates that there is no simple one to one rsFC and SC relationship possible between subjects, which is likely caused by individual differences in the brain. However, these results do provide a validation of the combined imaging approaches.
Next to the commonly known DMN and the above discussed thalamocortical system, nine RSNs have been identified in total so far. These RSNs includes the DMN, the core network, two lateralized parietal-frontal networks, primary motor, primary visual, extra-striate visual network and to singular networks consisting of bilateral medial frontal regions and posterior parietal cortical regions (Heuvel, van den, Mandl, Kahn and Hulshoff Pol 2009). Eight of these nine functionally linked RSNs seem to have a related underlying structural core. These results have been found in a resting state fMRI and DTI study by van den Heuvel et al. 2009. In this study they examined the existence of structural white matter bundles measured by DTI based fiber tracking between the functionally connected regions measured by resting state fMRI BOLD time-series of known RSNs as evidence for anatomical dependence of resting-state networks. They showed in 26 subjects that, with exception of the RSN consisting of the posterior precuneus, in all other well-known RSNs anatomical white matter tracts interconnected. This indicates that when looking to specific RSN in an within subject study, not only the DMN, but also other RSN networks do reflect the underlying structural connectivity architecture of the human brain.
When one examines multiple RSNs there seems to be a direct relationship between rsFC and SC (Heuvel et al. 2009). Furthermore, a review of a lot of individual studies show that the relationship between rsFC and SC relationship seems pretty clear (Damoiseaux and Greicius 2009). In the review by Damoiseaux and Greicius (2009), several rsFC and SC relationship studies where evaluated. They mostly compared direct rsFC and SC relationship studies in
respect to their difference in approach (whole brain versus ROI, mean FA versus number of tracts etc.), but also discussed clinical case studies. They concluded that rsFC correlates positively with SC, but also pointed out that rsFC was observed between regions where no SC was found. This probably indicates that functional connections are not only mediated by direct SC, but als by indirect SC. A functional connection between two brain regions could also be mediated by a third region instead due a direct structural connection between the two regions. But since DTI
techniques have an insufficient spatial resolution, these interpretations should be considered with caution.
The studies discussed above show that there is a positive relationship between rsFC and SC. This relationship is found within subjects as well as between subjects and is detectable in almost all commonly found RSNs. Overall there is strong evidence that the strength of rsFC is positively correlated with SC strength. However, functional connectivity is also observed between regions where there is little or no structural connectivity. Although in all studies a relationship between rsFC and SC is found, there still are some differences in the findings
between the divergent studies. To investigate to what extent these differences can be attributed to the design of the performed study, the next section will focus more on the used methods and how this possibly effects the findings.
Used Methods and Resting State Functional Connectivity and Brain Structure Relationships
Even though most of the results from studies discussed in the former section indicate a positive relationship between rsFC and SC, this relationship is not as straightforward as one might expect according to the Hebbian theory. Because the methods used in this kind of research are quite complicated and therefore might influence the findings, this section will focus on the used methods and how these methods possibly effect the findings. Damoiseaux and Greicius (2009) also compared methods in rsFC and SC studies, yet they focused more on differences in the technical aspects of MRI research. They discerned only small, negligible, differences in results between these different methods. To determine whether there is still a rsFC and SC relationship found if studies differ in other aspects of the design, five rsFC and SC relationship studies who differ in methodology will be assessed next.
Almost all rsFC and SC relationship studies focus on predetermined functional networks. Only Koch et al. (2002) took a more global approach by exploring rsFC in the entire brain, but even Koch’s study was limited to neighborhood gyri. This limitation probably explain the difference between the weak links Koch et al. (2002) found and the more strong correlations that where obtained from the global rsFC and SC comparisons from Skudlarski et al. (2008).
Skudlarski et al. (2008) used both DTI fiber tracking and resting state temporal correlations (RSTC) techniques to create global connectivity matrices covering the whole brain gray matter. Direct comparisons between rsFC measured by RSTC with anatomical connectivity quantified using DTI where made. They found that connectivity matrices obtained using both techniques showed significant agreement. A connectivity matrix is a graph or matrix format that represents brain connectivity (Sporns, 2007). The correspondence between these matrices was especially high in regions showing strong overall connectivity, such as regions belonging to the DMN.
To relate rsFC and SC it is also possible to take a step further than just using observed data and correlate them. If rsFC and SC are genuinely related, it should be possible to predict the one from the other. A very elegant way to test this hypothesis is performed by Honey et al. (2009). They used a computational model to test whether rsFC can be derived from SC. They used 5 participants and measured their resting state Functional Connectivity (fMRI) and
Structural Connectivity (DSI). First they compared these maps to each other, then they used the SC maps as couplings in a computational model. With this computational model they simulated both BOLD signals and functional connectivity, which then could be compared with the
empirical results. The results of this study indicate that when structural connections where present, there was a robust relationship between SC and rsFC in both the empirical data and the computational data. Although it seemed to be possible to predict strong rsFC from strong SC, the reverse inference was less reliable. Since this study simulated the rsFC in a computational model, it is more reliable than in the previously discussed studies to infer that the impossibility to predict SC from rsFC is caused by indirect connections instead of measurement artefacts. The use of a computational model makes therefore a valuable addition to the series of studies to rsFC and SC relationships.
Most of the studies in this field are done based on DTI measurements of white fiber tracts. Another way of assessing SC is by means of cortical thickness measurements, which gives
comparable results as white matter measurements in rsFC and SC relationship studies. He, Chen
and Evans (2007) investigated anatomical networks in the human brain by means of cortical thickness measurements from a MRI database. Two areas where considered to be anatomically connected if they had significant correlations of cortical thickness across the 124 investigated brains. They founded significant overlap between anatomical network modules and functional systems in the cortex. Thus, when changing the way of SC measurement from measuring white matter into measuring cortical thickness, a relationship between rsFC and SC still is found. However, it does make sense that when gray matter is also taken into account, a rsFC and SC relationship does not decreases, since BOLD is origin in grey matter.
Next to using a different method than DTI for SC measures, rsFC can also be measured in a different way than due to fMRI, namely due to EEG measurements. When combined with SC measures, EEG measurements for rsFC also appear to result in a relationship between rsFC and SC (Teipel et al., 2009). In their research study Teipel et al. 2009 investigated whether regional interhemispheric coherence determined using resting state EEG was associated with distributed networks of subcortical fiber tracts (DTI) both in healthy aging and in amnesic patients.
Interhemispheric coherence is an EEG technique for studying functional relationships between brain regions. Coherence of two EEG signals recorded from partially separated scalp electrodes estimates the similarity of waveform components generated by the summed action of neurons in underlying cortical regions (French and Beaumont, 1984). One of the waveforms that can be measured in EEG studies are α waves. The results of the experiment by Teipel et al. 2009 showed a pattern of predominant posterior white matter tracts associated with temporo-parietal α
coherence in amnesic patients and controls. In amnesic patients frontal α coherence was also related with integrity of predominantly frontal fiber tracts, yet in controls this relationship was not found. These results indicate that there is a relationship between rsFC and SC, but this relationship differs between amnesic patients and controls.
These studies indicate that the used method doesn’t really influence the finding of a relationship between rsFC and SC. Using a more global approach than specifically looking at RSNs, predicting rsFC from SC with a computational model, uses of cortical thickness instead of white matter measurements from DTI or alternate fMRI measures of rsFC with EEG measures of rsFC, the bottom-line still is the same; rsFC and SC are related. From most of the studies it does appear that rfFC can be predicted from SC but not vice versa, exactly in line with what has been found in the studies discussed in the previous paragraph. Thus the used method does not appear
to matter. The only aberrant result in the studies discussed above came from Teipel et al. 2009. Here a relationship in α coherence and frontal fiber tracts was found in amnesic patients but not in healthy controls. At first glance this does not seem such a remarkable result. A comparison is made between healthy and non-healthy brains, so a difference is to be expected. However, this result is obtained by something that has not been done in previous studies discussed, namely, by comparing different groups. This raises the question if a comparison of different groups could provide new information. Therefore the next paragraph will discuss studies of comparison in different groups with different features to investigate how rsFC relates to SC in (or between) these groups.
Variation in Functional and Structural Connectivity Relationships in Group Comparisons In general there seems to be a pretty robust relationship between rsFC and SC, yet this relationship might differ between different groups. The only comparison that has been made between groups with different features so far was between amnesic patients and a healthy control group. Here a difference in rsFC and SC relationships between these groups has been shown (Teipel et al., 2009). Even though it would be really interesting to investigate rsFC and SC relationships in pathological brains, for the purpose of this thesis comparisons between healthy groups are of greater importance. This because this thesis investigates a rsFC and SC relationship in general, and by studying pathological brains there is a risk to get bogged into details only applicable to a very small subgroup of the population. However, because of a lack of studies to individual differences in healthy subjects, some studies which investigate relatively common pathologies will also be discussed. If differences in healthy groups do exist, this could have important consequences for the interpretation of rsFC and SC relationship studies. Therefore this section investigates whether the previously found relationship between rsFC and SC remains intact in different groups. First, some studies with comparisons of healthy groups will be discussed. These groups include age and sex differences. Second, studies which contain comparisons of pathological brains will be treated, these include groups who suffer from brain injury, ADHD and schizophrenia.
One of the most obvious group differences one can think of when considering group comparisons in rsFC and SC relationship studies are age differences. Because children’s brains are still under development they differ significantly from adult brains, especially in respect to
white matter structure. Between birth and the third year a child the number of synapsis grows exponentially due to synaptogenesis. However, after the third year until adulthood this number of synapsis is halved due to synaptic pruning. Connections that are not used get pruned, and
connections that are used get strengthened (Chechik, Meilijson and Ruppin, 2006). From childhood to adulthood, there is ongoing white matter maturation in the brain. Hereby it seems pretty obvious that a difference in rsFC and SC relationships should exist. This inference is confirmed when comparing groups of children with adults, differences in rsFC and SC
relationships are usually found between these groups (Supekar et al. 2010, Uddin, Supekar, Ryali and Menon 2011). Although rsFC levels where similar in strength between these groups, SC in children was very weak or even not present at all. In contrast, underlying SC networks where shown in adults. Supekar et al. (2010) where the first to perform a study where white matter tracts where investigated in children by use of DTI tractography. They compared children (7-9 years old) with IQ matched young adults (19-22 years old) who both underwent resting state fMRI, DTI whole brain fiber tracking and VBM measurements. The DMN was the network of interest in this study. Results indicated that the DMN undergoes significant developmental changes in rsFC and SC relationships between childhood en adulthood. In childhood rsFC could reach adult-like levels, but there was weak SC. This strong rsFC in children might be found due to the high number of connections compared to adults. Although the functional networks do not have the underling efficient structural networks yet, there are enough connections to already provide the functional connectivity, which causes the maturation of the underlying SC resulting in SC networks in adulthood.
Uddin et al. (2011) examined developmental changes in brain network interactions using multimodal imaging, combining functional and effective connectivity (the union of structural and functional connectivity) of intrinsic fMRI and DTI tractography. Results have shown a significant correlation between rsFC and SC in adults, where in children this correlation was not present. The results of this study indicate that developmental changes in functional and effective connectivity seemed to be related to structural connectivity. These results are confirmed by Hagmann et al. (2010), who investigated these structural changes during development. Here a group of subjects in a range from 2 to 18 years is investigated to explored the contribution of white matter maturation to the development of connectivity using high b-value diffusion MRI tractography and connectivity analyses. Their results confirmed a positive correlation between
SC and rsFC, and in addition observed that this relationship strengthened with age. This
differences in rsFC and SC relationships between children and adults are not only interesting in the light of development, but also point out that a rsFC and SC relationship is not as robust and straightforward as most of the earlier discussed studies seemed to demonstrate. Especially the result that rsFC in children can have the same level of strength as in adults without any detectable underlying SC networks suggests this. This sets the whole view of rsFC and SC relationships in a different light, since rsFC doesn’t necessarily has to be the result of underlying SC. Although Damoiseaux and Greicius (2009) already pointed out that rsFC is also observed between regions where no SC is found, a whole network of activity without any underlying SC has not been observed earlier. This can be explained by previously discussed functional connections that are not only mediated by direct SC networks, but also by indirect connections. Still this would mean that a rsFC and SC relationship is not necessarily required for a functional RSN. But it is not clear what this lack of SC means for cognitive function. It also would be interesting to find out whether rsFC and SC relationships differ between age groups in adults. Since most studies are only performed in adolescent age groups, a differences between age groups would have
implications for the interpretations of rsFC and SC relationships in adults. Unfortunately there is no study that compares rsFC and SC in different age groups in adults yet.
Another class in which there seem to be differences in healthy subjects is gender (Filippi et al. 2013). In this study Filippi et al. 2013 sought to explore the impact on gender-related differences of rsFC of potential differences in brain morphology. A relatively large sample of healthy young adults in an narrow age range underwent both fMRI and VBM measurements of different RSNs. The aim of this study was to investigate differences and relationships in functional networks and cognitive functions. Since VBM is included as a measurement it is possible to infer some rsFC and SC relationships from the results. It appeared that some gender differences in function vanished after GM volumes used as a covariate, and some gender related differences in function were only detected after GM volume correction. This reveals rather ambivalent rsFC and SC relationship differences between genders. This study provides
insufficient information about rsFC and SC relationships to draw conclusions, but it is clear that gender differences do exist and should be taken into account while studying the brain.
Next to healthy brain comparisons also comparing groups who have disrupted SC networks with healthy subjects can reveal some information about rsFC and SC relationships.
Different pathologies seem to have different consequences for rsFC and SC relationships. In brain injuries and ADHD the disrupted SC is related to relatively low rsFC, which supports a direct relationship between rsFC and SC (Konrad and Eickhoff 2010, Sharp et al. 2011).
However, studies in schizophrenic patients also show an increase in rsFC in some disrupted SC systems (Skudlarski et al. 2010). Konrad and Eickhoff (2010) reported findings from rsFC studies, as well as from studies on SC using diffusion tensor imaging, in subjects with ADHD. According this a review by Konrad and Eickhoff 2010 there is convergent evidence for white matter pathology and disrupted anatomical connectivity in ADHD and related dysfunctional resting state connectivity has been demonstrated. Another study that shows a decrease in rsFC that comes together with SC disruption is from Sharp et al. (2011). They investigated whether rsFC is related to structural disconnection secondary to axonal injury (after traumatic brain injury), which they test by investigating whether rsFC correlates with abnormalities in adjacent white matter measured using DTI. Results showed that patients had widespread white matter damage compared with controls. Lower DMN FC was seen in those patients with more evidence of diffuse axonal injury within the adjacent corpus callosum. The results supported a direct relationship between white matter organization within the brain’s structural core and FC within the DMN. Which proves that white matter injury directly causes a disrupter in FC, which would be expected since FC and SC are related.
Although white matter tracts in schizophrenia patients are not injured, they do show disordered connectivity. Skudlarski et al. (2010) compared fMRI and DTI measures of
schizophrenia patients with a healthy control group. A global connectivity analysis indicated that patients had lower anatomical connectivity and lower coherence between the two imaging modalities. Although SC nearly uniformly decreased, FC in schizophrenia was lower for some connections (middle temporal gyrus) and higher for others (cingulate and thalamus). These results confirm the hypothesis that schizophrenia involve disordered connectivity between brain regions. It also is a clear example that a disrupted rsFC and SC relationship is associated with severe cognitive impairment.
When comparing rsFC and SC in different groups, a strong relationship is no longer uniformly found. In healthy subject comparisons, SC seems to develop over age. However this doesn’t apply for rsFC. As mentioned before this is probably due to white matter maturation. This indicates that rsFC is not always in the same extent related to SC. Also gender comparisons
provide a rather ambivalent relationship between rsFC and SC. To make this even more complex also rsFC and SC relationships comparison studies in pathological brains show contrary results. In subjects with white matter injuries and in ADHD patients the relationship between rsFC and SC seemed to be pretty straightforward again (both lower rsFC and SC) but in schizophrenia patients there didn’t appears to be a clear relationship anymore.
Conclusion and discussion
RsFC can be predicted from SC, however, the reverse inference is less reliable. There seems to be a positive relationship between rsFC and SC in most RSNs. Although the strength of rsFC is positively correlated with SC strength, FC is also observed between regions where there is little or no SC. The used method doesn’t seem to have a big influence to this relationship, even with big methodological adjustments a basic rsFC and SC relationship is still found. One factor that does seem to have influence on a founded relationship between rsFC and SC its individual differences. When comparing healthy adults by with children, the relationship between rsFC and SC seems to disappear. SC develops over age, but rsFC remains the same. Additional, gender differences seem to have a strange influence on a rsFC and SC relationship. When comparing healthy adults with groups who suffer from mental health issues, rsFC and SC relationships are confirmed in some conditions, but seem to disappear in other. Taken all together this seems to indicate that although rsFC doesn’t always reflects SC, SC mostly seems to produce FC. RsFC can be predicted from SC, but only with some certainty in healthy adults within a narrow age range. RsFC does not always have to come from SC, also indirect connectivity exists. This is probably caused by indirect connections, which means that a region in a functional network might be activated by a third region in some cases, instead of activation due to direct underlying SC.
Even though this review only asses rsFC related to SC other functional measures could be very valuable to add to the investigation. Especially with respect to individual differences in rsFC and SC relationships. By, for example, using cognitive measures to add to the comparison of rsFC and SC, you might be able to indicate whether if rsFC and SC relationships decreases, it accompanies with a decrease in cognitive function. This would give a more clear sight on what happens when direct rsFC and SC relationships are not present. For example, in children it would
be very interesting to test cognitive functions associated with the DMN. Hereby it would be possible to test where their cognitive function differs from adults. These differences can probably be specifically attributed to the SC differences, since FC is mostly the same between adults and children.
Beside taking the cognitive aspect into account the use of computational models can also provide a lot more information about how brain function and structure exactly relate to each other. The use of a computational model for rsFC simulations could, for example, provide evidence for indirect connectivity in the brain. These indirect connections can only be inferred from the results the studies discussed in this thesis showed. By modelling indirect connectivity, it would be possible to see if this indirect connectivity arises in the model en how these connections run through the brain.
Another aspect to keep in mind within future research is the fact that group variance can have a big influence on the relationship between rsFC and SC. This should be more important focal point in rsFC and SC relation studies. Although most of the studies where performed with healthy subjects in narrow age groups, most of the studies that investigated a relationship between rsFC and SC didn’t explicitly included this point in their design. Future research it should take this into account, by, for example, give a clear description of how they include variables as age and sex differences in their analyses.
With a growing number of new techniques to measure brain function as well as brain structure, the past years have been providing a lot of new knowledge about brain function an structure relationships. Together with this knowledge however, many new questions have also came up. Hopefully these new insights also provide new clues and techniques to make the questions of today, the answers of the future.
Literature
Chechik, G., Meilijson, I., & Ruppin, E. (2006). Neuronal Regulation: A Mechanism for Synaptic Pruning During Brain Maturation, Neural Computation, 11(8), 2061-2080.
Damoiseaux, J.S., & Greicius, M.D. (2009). Greater than the sum of its parts: a review of studies combining structural connectivity and resting-state functional connectivity, Brain
Structure and Function, 213(6), 525-533.
Filippi, M., Valsasina, P., Misci, P., Falini, A., Comi, G., & Rocca, M.A. (2013). The organization of intrinsic brain activity differs between genders: A resting-state fMRI study in a large cohort of young healthy subjects, Human Brain Mapping, 34(6), 1330-1343.
French, C.C., & Beaumont, J.G. (1984). A critical review of EEG coherence studies of hemisphere function, International Journal of Psychophysiology, 1(3), 241-254. Greicius, M.D., Supekar, K., Menon, V., & Dougherty, R.F. (2009). Resting-State Functional
Connectivity Reflects Structural Connectivity in the Default Mode Network, Cerebral Cortex, 19(1), 72-78.
Hagmann, P., Sporns, O., Madan, N., Cammoun, L., Pienaar, R., Wedeen, V.J., Meuli, R., Thiran, J.-P., & Grant, P.E. (2010). White matter maturation reshapes structural connectivity in the late developing human brain, PNAS, 107(44), 19067-19072. He, Y., Chen, Z.J., & Evans, A.C. (2007). Small-World Anatomical Networks in the Human
Brain Revealed by Cortical Thickness from MRI, Cerebral Cortex, 17(10), 2407-2419. Heuvel, M.P. van den, Mandl, R.C.W., Kahn, R.S., & Hulshoff Pol, H.E. (2009). Functionally
linked resting-state networks reflect the underlying structural connectivity architecture of the human brain, Human Brain Mapping, 30(10), 3127-3141.
Heuvel, M.P. van den, & Hulshoff Pol, H.E. (2010). Exploring the brain network: A review on resting-state fMRI functional connectivity, European Neuropsychopharmacology, 20(8), 519-534.
Honey, C.J., Sporns, O., Cammoun, L., Gigandet, X., Thiran, J.P., Meuli, R., & Hagmann, P. (2009). Predicting human resting-state functional connectivity from structural
connectivity, PNAS, 106(6), 2035-2040.
Koch, M.A., Norrit, D.G., & Hund-Georgiadis, M. (2002). An Investigation of Functional and Anatomical Connectivity Using Magnetic Resonance Imaging, NeuroImage, 16(1), 241-250.
Konrad, K., & Eickhoff, S.B. (2010). Is the ADHD brain wired differently? A review on structural and functional connectivity in attention deficit hyperactivity disorder, Human Brain Mapping, 31(6), 904-916.
Kwong, K.K., Belliveau, J.W., Chesler, D.A., Goldberg, I.E., Weisskoff, R.M., Poncelet, B.P., Kennedy, D.N., Hoppel, B.E., Cohen, M.S., & Turner, R. (1992). Dynamic magnetic resonance imaging of human brain activity during primary sensory stimulation. PNAS, 89(12), 5675-5679.
Le Bihan, D. (2003). Looking into the functional architecture of the brain with diffusion MRI, Nature Reviews Neuroscience, 4, 469-480.
Raichle, M.E., MacLeod, A.M., Snyder, A.Z., Powers, W.J., Gusnard, D.A., & Shulman, G.L. (2001). A default mode of brain function, PNAS , 98(2), 676-682.
Sharp, D.J., Beckmann, C.F., Greenwood, R., Kinnunen, K.M., Bonnelle, V., Boissezon, X. De, Powell, J.H., Counsell, S.J., Patel, M.C., & Leech, R. (2011). Default mode network functional and structural connectivity after traumatic brain injury, Brain, 134(8), 2233-2247.
Skudlarski, P., Jagannathan, K., Calhoun, V.D., Hampson, M., Skudlarska, B.A., & Pearlson, G. (2008). Measuring brain connectivity: Diffusion tensor imaging validates resting state temporal correlations, NeuroImage, 43(3), 554-561.
Skudlarski, P., Jagannathan, K., Anderson, K., Stevens, M.C., Vince D. Calhoun, V.D., Skudlarska, B.A., & Pearlson, G. (2010). Brain Connectivity Is Not Only Lower but Different in Schizophrenia: A Combined Anatomical and Functional Approach, Biological Psychiatry, 68(1), 61-69.
Sporns, O. (2007) Brain connectivity. Scholarpedia, 2(10):4695.
Supekar, K., Uddin, L.Q., Prater, K., Amin, H., Greicius, M.D., & Menon, V. (2010).
Development of functional and structural connectivity within the default mode network in young children, NeuroImage, 52(1), 290-301.
Suzuki, W.L., & Amaral, D.G. (1994). Perirhinal and parahippocampal cortices of the macaque monkey: Cortical afferents, Journal of Comparative Neurology, 350(4), 497-533.
Teipel, S.J., Pogarell, O., Meindl, T., Dietrich, O., Sydykova, D., Hunklinger, U., Georgii, B., Mulert, C., Reiser, M.F., Möller, H-J., & Hampel, H. (2009). Regional networks
underlying interhemispheric connectivity: An EEG and DTI study in healthy ageing and amnestic mild cognitive impairment, Human Brain Mapping, 30(7), 2098-2119.
Uddin, L.Q., Supekar, K.S., Ryali, S., & Menon, V. (2011). Dynamic Reconfiguration of
Structural and Functional Connectivity Across Core Neurocognitive Brain Networks with Development, The Journal of Neuroscience, 31(50), 18578-18589.
Zhang, D., Snyder, A.Z, Shimony, J.S., Fox, M.D., & Raichle, M.E. (2010). Noninvasive Functional and Structural Connectivity Mapping of the Human Thalamocortical System, Cerebral Cortex, 20(5), 1187-1194.
Research Proposal
Can Structural and Functional Connectivity properties be revealed by combining differences in these measures with cognitive measures?
Name: Sabrina Okx
Student Number: 10003422 Date: 05-06-2014
Supervisor: Olympia Colizoli Number of Words: 1854
Abstract
A relatively new but already well investigated topic in Cognitive Neuroscience is the relationship between rsFC and SC. Although this relationship itself has been researched extensively, so far
nobody has analyzed how this structure function relationship is related to cognitive measures. This study will therefore explore if it is possible to reveal Structural and Functional Connectivity
properties by combining differences in these measures with cognitive measures. First the results of a former study are tried to be replicated, by comparing resting state Functional Connectivity (rsFC) and Structural Connectivity (SC) between the Posterior Cingulate Cortex and the Medial Temporal Lobes of adults and children. It is expected that children will have the same high levels
of rsFC as adults in this area, but that the underlying SC is lower in children than in Adults. Second these rsFC and SC measures are related to Episodic Memory Scores, to see how cognitive
function relates to the rsFC and SC relationship.
Functional and Structural Connectivity Relationships and Episodic Memory Function in Children and Adults
One of the latest trends in Neuroimaging studies is examining the relationship between resting state Functional Connectivity (rsFC) and Structural Connectivity (SC). RsFC is measured with functional Magnetic Resonance Imaging (fMRI), which is a neuroimaging technique based on Magnetic Resonance Imaging (MRI), where oxygenated blood flow is measured. Since oxygenated blood flow in the brain is related to neuronal activation, measuring this blood flow makes it possible to obtain completely non-invasive maps of human brain activity. Resting state fMRI studies examine functional interactions between brain regions by correlating the measured spontaneous brain activity. This rsFC is suggested to describe the relationship between the neuronal activation patterns of anatomically separated brain regions, reflecting the level of functional communication between regions (van den Heuvel and Hulshoff Pol, 2010). Brain regions that show synchronized activity are said to form resting-state networks (RSNs) (van den Heuvel et al. 2009). The high level of resting state Functional Connectivity (rsFC) within these RSNs suggests that the involved brain areas are not only functional linked, but that there is an underlying structural pathway that provides this observed activity.
To investigate these suggested underlying structural pathways, rsFC is related to SC, which is measured with DTI. DTI is a MRI technique that measures the diffusion of water
molecules in the brain (Le Bihan 2003). In the brain, water diffuses more promptly along axons than across them, and measuring the direction of this diffusion can therefore be used to infer the orientation of white matter tracts in the brain. By relating these correlated fMRI measures (rsFC) with DTI (SC) measures, several studies have found that there is a positive relationship between rsFC and SC (Damoiseaux and Greicius 2009). Overall there is strong evidence that the strength of rsFC is positively correlated with SC strength. However, functional connectivity is also observed between regions where there is little or no structural connectivity.
A remarkable finding in this line of research is a difference in rsFC and SC relationship between children and adults. Supekar et al. (2010) were the first to perform a study where white matter tracts where investigated in children by use of DTI tractography. They compared children (7-9 years old) with IQ matched young adults (19-22 years old) who both underwent resting state fMRI and DTI whole brain fiber tracking measurements. The Default Mode Network (DMN) was the network of interest in this study. The DMN is a RSN that covers a number of frontal,
temporal and parietal areas of the brain that shows high rsFC and typically deactivates during performance of cognitive tasks (Raichle et al. 2001). One of the most interesting findings of Supekar et al. (2010) was that some parts of the DMN in children could reach adult-like levels of rsFC but the underlying SC was weak, whilst in adults the rsFC in these areas was related to the underlying SC. Since the rsFC in some parts of the DMN is the same for children and adults but the SC is not, it would be interesting to compare these findings with measurements of cognitive functions related to these areas. If these cognitive measures are different between children and adults this difference is probably related to the difference in SC (since rsFC is the same).
However, if these cognitive measures are the same in children as in adults this indicates that rsFC mainly defines cognitive function, regardless of the underlying SC. To investigate how the rsFC and SC relationship is related to cognitive measures, this study tries to explore if it is possible to reveal Structural and Functional Connectivity properties by combining differences in these measures with cognitive measures.
First of all this study will try to replicate the findings of Supekar et al. (2010), by examine rsFC and SC in children and adults. Because the difference in rsFC and SC relationship between children and adults was clearest between the Posterior Cingulate Cortex and Medial Temporal Lobes (Supekar et al. 2010), this study will specifically focus on this part of the DMN. A group of children as well as a group of adults will undergo a MRI scanning session, where rsFC
(correlated resting state fMRI activity) and SC (DTI tractography measures) will be measured. It is expected that findings will be in line with the findings of Supekar et al. (2010). Namely, that rsFC between Posterior Cingulate Cortex and Medial Temporal Lobes reaches the same levels in children as in adults and white matter density between Posterior Cingulate Cortex and Medial Temporal Lobes will be lower in children than in adults.
In addition this study examines how a cognitive measure is related to the rsFC and SC differences between children and adults. Because the Posterior Cingulate Cortex and Medial Temporal Lobes connections in the DMN are mostly related to Episodic Memory (Staresina and Davachi 2008), Episodic Memory will be used as the related cognitive measure in this study. Episodic Memory is the memory of autobiographical events, consisting of semantic as well as contextual properties (Helm, Gujar, Nishida and Walker 2011). To measure Episodic Memory an Episodic Memory test developed by Helm et al. (2011) is used. This test includes both the
semantic and the contextual aspects of a typical Episodic Memory. Since earlier studies of Episodic Memory in children demonstrated that Episodic Memory is almost fully developed by the age of six (Marshall, Drummey, Fox and Newcombe 2002), it is expected that there will be no difference between Episodic Memory results in children and adults.
Research Methods Subjects
- Twenty children between 9 and 10 years old, all right handed females with a Dutch nationality. - Twenty young adults between 19 and 22 years old, , all right handed females with a Dutch nationality. All adults are IQ-matched with the children.
Scan
- MRI Philips 3T Achieva.
- Software for acquiring functional and DTI tractography measures. Task
- Laptop with Presentation software.
- Episodic memory task as developed by Staresina and Davachi (2008), consisting of 700 three to eight letter long nouns.
Procedure
First all participants are selected by a Raven´s IQ test, as well as demographic values like
age and handedness. Second the selected participants will perform the encoding/learning part of the episodic memory task. Here participants are shown 420 nouns in random order, in 4 runs each consisting of 105 words. The words are presented in a colored square (blue, green red or yellow) and the participants are instructed to vividly imagine the referent of the noun in the background color and to decide whether this combination is plausible or appealing. Participants can press “Z” ate the laptop keyboard for plausible and “M” for appealing. If participants cannot imagine the noun/color combination properly they have to press the spacebar, these items will be excluded from further analyses.
After finishing this task participants will be prepared for the scan session. All subjects undergo an 8-minute long resting state scan, whilst being instructed to close their eyes and relax. When the scan session is finished, approximately 30 minutes after finishing the first part of the episodic memory task, participants have to perform a surprise recognition memory task. The 420 previously shown nouns are presented randomly alternated with 280 new words. Participants are asked to indicate due to a button press if the presented word is “new” or “old”. When the answer is “old”, multiple choice question appear to ask the color and to ask how the item has been rated (plausible vs appealing). In both questions participants can also choose a question mark if they do not know the answer.
Data Analysis Scan session
- ROIs are anatomically selected in the Posterior Cingulate Cortex and the Medial Temporal Lobes in both hemispheres.
- Regional resting state fMRI time series is computed and partial correlation is used as a rsFC strength measure.
- Fiber density is taken as a measure for Structural Connectivity, which is computed from DTI tractography measurements by the formula described in Supekar et al. (2010).
Comparisons of rsFC and SC in children and adults
- RsFC between Posterior Cingulate Cortex and the Medial Temporal Lobes is characterized using ICA.
- RsFC and fiber density of adults and children is compared due a two sample t-test. - RsFC and SC relationship is calculated by a Pearson Correlation.
Episodic Memory Task
- The mean number of correct recognized words is compared for adults vs children due to an independent ANOVA.
Comparisons of Scan results and Episodic Memory Scores
- Relate Episodic Memory Score with rsFC and SC measures by a Pearson Correlation
Interpretation of Possible Results
Scanning results are expected to be the same as in the study of Supekar et al. 2010, namely, lower SC in Posterior Cingulate Cortex and Medial Temporal Lobes in children than in adults, but the same level of rsFC between these areas in children and adults. Episodic Memory results are expected to be the same for children as for adults. Because this study aims to replicate the rsFC and SC relationship difference between children and adults, only the Episodic Memory relationships to this outcome will be discussed here. Other possible results can also be interesting, but that is too extensive to discuss here.
When rsFC is the same in children and adults and SC is lower in children than in adults: - If mean Episodic Memory scores in children are worse than in adults (e.g. they make more mistakes in the old/new judgment), this will mean that SC plays an important role in Episodic Memory function.
- If mean Episodic Memory scores are the same in children as in adults (e.g. they make the same amount of mistakes in the old/new judgment), this will mean that rsFC determines Episodic Memory function.
- If mean Episodic Memory scores in children are better than in adults (e.g. they make less mistakes in the old/new judgment), this will mean that SC is negatively related to Episodic Memory function.
Note: Whit the interpretation of these results you should always keep in mind that other brain regions/networks also play a role making the Episodic Memory test. This can make it hard to make a clear interpretation of these results.
Literature
Damoiseaux, J.S., & Greicius, M.D. (2009). Greater than the sum of its parts: a review of studies combining structural connectivity and resting-state functional connectivity, Brain
Structure and Function, 213(6), 525-533.
Greicius, M.D., Srivastava, G., Reiss, A.L., & Menon, V. (2004). Default-mode network activity distinguishes Alzheimer's disease from healthy aging: Evidence from functional MRI, PNAS, 101(13), 4637-4642.
Helm, E., van der, Gujar, N., Nishida, M., & Walker, M.P. (2011). Sleep-Dependent Facilitation of Episodic Memory Details, PLoS One, 6(11), e27421.
Heuvel, M.P. van den, Mandl, R.C.W., Kahn, R.S., & Hulshoff Pol, H.E. (2009). Functionally linked resting-state networks reflect the underlying structural connectivity architecture of the human brain, Human Brain Mapping, 30(10), 3127-3141.
Heuvel, M.P. van den, & Hulshoff Pol, H.E. (2010). Exploring the brain network: A review on resting-state fMRI functional connectivity, European Neuropsychopharmacology, 20(8), 519-534.
Le Bihan, D. (2003). Looking into the functional architecture of the brain with diffusion MRI, Nature Reviews Neuroscience, 4, 469-480.
Marshall, D.H., Drummey, A.B., Fox, A.A., & Newcombe, N.S. (2002). An Event-Related Potential Study of Item Recognition Memory in Children and Adults , Journal of Cognition and Development, 3(2), 201-224.
Raichle, M.E., MacLeod, A.M., Snyder, A.Z., Powers, W.J., Gusnard, D.A., & Shulman, G.L. (2001). A default mode of brain function, PNAS , 98(2), 676-682.
Staresina, B.P., & Davachi, L. (2008). Selective and shared contributions of the hippocampus and perirhinal cortex to episodic item and associative encoding, Journal of cognitive
neuroscience, 20(8), 1478-1489.
Supekar, K., Uddin, L.Q., Prater, K., Amin, H., Greicius, M.D., & Menon, V. (2010).
Development of functional and structural connectivity within the default mode network in young children, NeuroImage, 52(1), 290-301.
