In addition, each WM-FN was divided into left and right sides for subsequent DTI analysis. For each subject, fiber bundles connecting the left and right sides of each WM-FN through corpus callosum were extracted from the whole-brain fibers, which was implemented in the TrackVis

The ten WM-FNs described above were selected as ROIs for DTI analysis

re the fibers connecting bilateral WM-FNs through the callosal region were tracked, these WM-FNs and the corpus callosum mask of MNI space (voxel size: 3 × 3 × 3 mm3) were resampled to standard MNI diffusion space with a voxel size of 1.25 × 1.25 × 1.25 mm3 using the REST software package

Clustering white-matter functional networks

The corpus callosum mask

To obtain WM-FNs, we created the group-level WM mask. Specifically, the T1-weight images were segmented into GM, WM, and CSF using the DPARSF software based on SPM12. Each voxel of whole brain was identified as GM, WM, and CSF using the maximum probability from the segmentation images, which generated three masks of GM, WM, and CSF for each subject. Subsequently, we averaged these high-resolution anatomical segmented WM masks across all subjects, and the group-level WM mask was defined by using a threshold greater than 60% of the subjects

Data Acquisition

The HCP data for resting state fMRI had four runs (rfMRI_rest1_LR, rfMRI_rest1_RL, rfMRI_rest2_LR, and rfMRI_rest2_RL).

(subjects were excluded because DTI data was not collected (4 subjects) or because of different DTI scan parameters (7 subjects)).

Subsequently, the partial correlation analysis was calculated between WM-FNs and each callosal voxel time courses. Then a statistical correction (p < 0.05, FWE-corrected) was performed on the one-sample t-test map of each WM-FN as a region of interest (ROI). In this end, to assess the significance of overlap between DTI and RSFC results, Monte Carlo Calculation was performed in the study.

However, our previous study did not analyze the structural connectivity between the corpus callosum and WM-FNs, nor did it map the correspondence between the structural and functional connectivity pertaining to the corpus callosum.

More importantly, the organization of corpus callosum was divided into ten functional subregions corresponding to distinct WM-FNs using RSFC in our previous study

The topographical distribution of fiber bundles connecting to the cortex has been established in midsagittal cross-sections of the corpus callosum using a diffusion MRI-based parcellation method (Huang et al., 2005; Hofer and Frahm, 2006; Pannek et al., 2010; Archer et al., 2019).

Moreover, several studies have also analyzed the underlying BOLD fMRI signals in the corpus callosum and topographical organization of corpus callosum corresponding to human primary and higher-order functional systems. Distinct areas of the corpus callosum were found to have distinct activation patterns in response to the different tasks, such as motor, visual, tactile, gustatory, memory, and auditory tasks

Our previous study explored the functional connectivity between the corpus callosum and white-matter functional networks (WM-FNs)

By performing K-means clustering on the voxel-based WM functional connectivity matrix, Peer and colleagues obtained distinct symmetric WM-FNs, which were closely associated with both GM-FNs and the underlying structural WM tracts (Peer et al., 2017). Using test-retest data from the Human Connectome Project (HCP), our previous study has also found reproducible WM-FNs, as well as functional connectivity between WM- and GM-FNs (Wang et al., 2020).

These findings reinforce the notion that the corpus callosum has unique spatial distribution patterns connecting to distinct WM-FNs.

esults demonstrated that the structural and functional connectivity between the human corpus callosum and WM-FNs have the following high overall correspondence:

uses the resting-state fMRI of Human Connectome Project data to identify ten WM-FNs

corresponding physical connectivity remains unknown

Lastly, the number of connected voxels in the callosal midsection showed a decrease at both ends of the gradient, while gradient values in the middle appeared more similar.

Mapping the principal gradient onto the corpus callosum

Another limitation comes from the tractography approach. Tractography analyses can produce inaccurate results (Jones and Cercignani, 2010; Maier-Hein et al., 2017). To avoid these caveats, we employed methods that have previously demonstrated high anatomical reliability when compared to axonal tracing (Thiebaut de Schotten et al. 2011a; 2012) and with post-mortem dissections (Catani and Thiebaut de Schotten, 2012; Thiebaut de Schotten et al., 2011, 2017; Vergani et al., 2014).

However, this result is not generalizable to all cognitive functions as for example studies investigating speech processes (i.e.: word fluency and semantic decision making) indicate that a ‘stronger’ corpus callosum increases asymmetries between the two hemispheres

structural variability is associated with evolutionary stability

As such, the callosal gradient may putatively be related to different conduction speeds, which needs to be addressed by future research.

distribution of values might be related to a wider range of neuropsychological syndromes associated with disconnections of the same regions of the corpus callosum

With some regions (3,2, 1) being highly consistent in terms of gradient values and region 7 to 4 being more distributed

region previously shown as connecting the angular gyrus (Park et al., 2008; Chao et al., 2009) well known for its high values in the principal gradient (Margulies et al., 2016) and central structure of the default mode network

Gradient values in the splenium were heterogeneous with highest values

genu connects to higher-level associative regions

esearch demonstrated that the posterior midbody contains fibers that project to lower-level processing areas

The callosal gradient (Fig. 2E, left) indicates that the negative spectrum of the principal gradient peaks in the posterior body, whereas the positive spectrum peaks in the genu.

Previous research investigated the topography of callosal connections solely based on atlases that parcellate the cortex into areas (Chao et al., 2009) or gyri (Park et al., 2008) that are not defined by their functional role nor their hierarchy in the brain. Our study circumvents this limitation by utilizing the principal gradient that has been linked to functional hierarchy

cond, the individual variability of callosal connectivity is lowest at the apexes of the gradien

First, the apexes of the principal gradient span from the genu to the posterior body.

Similar to the previous analysis, a curve-fit estimation was employed to investigate a linear or quadratic relationship between the number of callosal voxels that may connect to a certain gradient value.

Relationship between the principal gradient and callosal connectivity

Results indicate a structured distribution of the principal gradient along the midsection of the corpus callosum where the anterior corpus callosum is more likely connected to areas of higher principal gradient values compared to posterior areas.

Mapping the principal gradient onto the corpus callosum

We manually delineated the corpus callosum midsection on the midsagittal slice of the MNI152 template.

In this study we propose a novel topography of the corpus callosum based on the projection of the principal gradient onto the callosal midsection. We also explored how variability in connectivity and quantitative measures of this connectivity relate to the principal gradient values along the callosal midsection.

All these anatomical, histological and geometrical parcellations of the midsection of the corpus callosum do however not consider the functional organisation of the brain areas it connects.

tentatively

Concordantly

With 200 to 300 millions axons (Tomasch 1954), the largest white matter commissure in the primate brain is the corpus callosum (Burdach, 1819).

While this principal gradient is putatively the same for both hemispheres (Margulies et al., 2016) little is known on the connections between the left and the right principal gradient.

For instance, Mesulam (1998) proposes that brain regions can be organised along a gradient ranging from sensory-motor to higher-order brain processes. I

adheres

principal gradient

Resting-state functional connectivity (rsfMRI), in particular, has proven useful for mapping multiple functional networks without relying on participants to perform functional tasks in the scanner

Cognitive neuroscience assumes that cognition is embedded within and constraint by the organization of the brain. The advent of neuroimaging methods granted novel insights into the functional organization of the living human brain.

Our results shed light on the relationship between the brain's functional hierarchy and the corpus callosum

In so doing, we demonstrated that the extreme values of the principal gradient are located within the callosal genu and the posterior body, have lower connectivity variability but a larger spatial extent along the midsection of the corpus callosum than mid-range values.

. While the cortex has been well characterised in terms of gradients little is known about its underlying white matter.

General Conclusion

Conclusion

Task-Related Studies

Resting State Studies

Source-Level Analysis

All the reviewed studies support the notion that structural connectivity is associated with both task-related and resting state functional coupling at certain frequency bands. Overall, it seems that reduced coherence/phase synchronization is accompanied by an altered WM microstructural organization.

Task-Related Studies

Resting State Studies

Synchronization Analysis

The strength of functional connectivity may depend on the microstructural organization of neuronal fibers crossing through the cerebral WM

A set of functionally connected regions is referred to as a functional network. Some functional networks can be detected during the resting state (Teipel et al., 2009), while others in the context of task-related behavior (Cohen, 2011; Liu et al., 2017) some of which persist across behavioral states like the default mode network

Conclusion

For this, we speculate that the increased delta and theta power were correlated with decreased FA values.

They found a significant correlation between FA values and delta/theta power metrics (averaged over electrodes) for patients and age-matched healthy controls. In healthy controls, delta power was correlated with FA values in the left superior corona radiata (connecting medial frontal cortex and basal ganglia), which was absent in the patients. The theta power was correlated with FA values in the body of corpus callosum, the left cingulum, and in bilateral superior longitudinal fasciculi in the healthy controls, and in patients in the right external capsule

In the left frontal WM and in the corpus callosum, the progressive decrease in FA values appeared to be significantly correlated with increased delta and theta rhythm in frontopolar and frontal ROIs.

Spectral Analysis – Local Functional Connectivity

ERP components can hence be considered sensitive to microstructural differences and, in consequence, amplitudes and latencies of ERP components may reflect structural changes in pathology during cognitive, sensory, or motor performance.

velocity

Effects of structural connectivity on functional activity may explain differences in amplitudes and latencies of ERP components, both in healthy participants and patients

Reduced mismatch responses (less-negative amplitudes) and reduced FA of the auditory interhemispheric pathway were associated with psychotic-like experiences. Both of them may hence marker early psychotic symptoms.

The results hence suggest that P300 latencies can be affected by microstructural organization of the corpus callosum. Put differently, the microstructural organization of the corpus callosum seems to correlate with cognitive functions

also reported that FA values were negatively correlated with a latency period of the P300 in the anterior quadrant of the epileptogenic hemisphere (decreased FA values correlated with a longer P300 latencies).

They found a significant, negative correlation between epileptic activity and FA values, in which increased amplitude of theta rhythm correlated with decreased FA values in the hemisphere with epileptic activity.

They observed a significant positive correlation between the corresponding FA values and the degree of N1 suppression in the un-delayed condition (higher FA values correlated with higher N1 suppression) and a negative correlation with N1 suppression in one of the delayed conditions (50 ms).

. They found an association between longer IHTT (P100 latency differences) and lower FA values in visual callosal fibers for both groups, albeit not significant after Bonferroni correction.

To further detail spatial dependencies, Whitford et al. (2011a) applied tractography between selected ROIs. They investigated the correlation between IHTT and structural connectivity but concentrated on fibers of the corpus callosum connecting visual processing areas in groups of schizophrenia patients.

While this seemingly contradicts the aforementioned mechanistic account, it does suggest a direct relationship between microstructural organization of the posterior regions of the corpus callosum and inter-hemispheric transfer time of visual information.

This revealed significant correlations between longer left/right occipital P100 IHTT and lower MD values in callosal regions connecting visual cortical areas.

As compared to the collar group, controls showed larger electrophysiological changes along with increased MD/RD values from pre- to mid-season in the corpus callosum.

Myer et al. (2016) evaluated the so-called Q-Collar, a neck collar worn by athletes with the intention to protect the brain from head impacts. Aim of this study was to test the collar’s effect on reducing neuroanatomical and neurophysiological damage in two groups of hockey players that either wore the collar or did not (the latter served as controls).

With this, Gao et al. (2017) found a significant positive correlation between larger N2 amplitudes and higher FA in right superior longitudinal fasciculus connecting frontal and parietal cortex.

determined

They also reported a negative correlation between the ERN amplitude size and RD values, i.e., smaller RD values correlated with more negative ERN amplitudes in the left cingulum bundle.

found a positive correlation between ERN amplitudes over medial fronto-central EEG electrodes and mean FA values in left posterior cingulate cortex and left superior longitudinal fasciculus.

Cognition-related responses are later ERPs resulting from attention and cognitive control processes, e.g., when participants respond to stimuli like words or images

Early attention-related responses can be observed when the task is to react (respond) as soon as an external sensory stimulus has been observed

We grouped them under the headings error-related, early attention-related and later cognitive-related responses

Error-related responses are ERPs evoked by commission errors when participants are asked to respond as accurately and quickly as possible to a cognitive task

All the reports included in this section are so-called task-related studies, where participants were asked to perform a cognitive, sensory, or motor task during EEG recording.

Especially FA is believed to reflect this degree of WM microstructure and should therefore be associated with different outcome measures like ERP amplitudes and latencies

An increased degree of microstructural organization correlates with faster action potential conduction

The most used approach in experimental EEG studies comprises of evoking activity patterns using repeated stimuli that may differ between studies. This design yields so-called event-related potentials (ERPs), highlights activity that is time-locked to an event (i.e., the moment of stimulation) and eliminates non-time-locked background activity via averaging to event-aligned epochs. According to the aforementioned mechanistic view (see Section “A Mechanistic View”), the possibility of time-locked activity to emerge depends on the myelination of connecting axons and dendrites

refrained

selecting the proper statistical means to identify possible associations between them.

defining appropriate (statistical) methods to determine the structural and functional characteristics of interest

Estimating the relationship between structural and functional brain characteristics requires three important steps:

Literature Review

A Mechanistic View

(i) selecting a set of ROIs, either a priori or based on the statistical significance of FA group differences or signal power as a function of location

Lastly, KA solely reflects the anisotropy of the kurtosis tensor without contributions from the diffusion tensor (Poot et al., 2010). All the measures can be estimated in isolated ROIs or for fibers connecting them, the latter after conducting additional tractography; see Figure 1.

Introduction

Many studies followed this idea and combined dMRI with fMRI, in parts with great success (Mulkern et al., 2006; Johansen-Berg, 2012; Bennett and Rypma, 2013).

Functional connectivity refers to statistical dependence between time series of electro-physiological activity and (de)oxygenated blood levels in distinct regions of the brain.

It appears that the strength of event-related EEG activity increases with increased integrity of structural connectivity, while latency drops.

glance

White matter microstructural organization and functional activity can be assessed in unison.

counterpart

Implications