Michael W.L. Chee

Sleep deprivation reduces default mode network connectivity and anti-correlation during rest and task performance

Sleep deprivation (SD) can alter extrinsic, task-related fMRI signal involved in attention, memory and executive function. However, its effects on intrinsic low-frequency connectivity within the Default Mode Network (DMN) and its related anti-correlated network (ACN) have not been well characterized. We investigated the effect of SD on functional connectivity within the DMN, and on DMN-ACN anti-correlation, both during the resting state and during performance of a visual attention task (VAT). 26 healthy participants underwent fMRI twice: once after a normal night of sleep in rested wakefulness (RW) and once following approximately 24h of total SD. A seed-based approach was used to examine pairwise correlations of low-frequency fMRI signal across different nodes in each state. SD was associated with significant selective reductions in DMN functional connectivity and DMN-ACN anti-correlation. This was congruent across resting state and VAT analyses, suggesting that SD induces a robust alteration in the intrinsic connectivity within and between these networks.

Dissociation of Cortical Regions Modulated by Both Working Memory Load and Sleep Deprivation and by Sleep Deprivation Alone

Working memory is an important mental capacity that is compromised following sleep deprivation (SD). To understand how working memory load interacts with state to influence brain activation in load-sensitive regions, and the extent to which SD-related changes are common across different loads, we used fMRI to study twelve healthy subjects following 24 h of SD using a verbal n-back task with three load levels. Performance decline was observed by way of reduced accuracy and slower response times following SD. The left prefrontal region and thalamus showed load dependent activity modulation that interacted with state. The right parietal and anterior medial frontal regions showed load dependent changes in activity as well as an effect of state. The anterior cingulate and occipital regions showed activation that displayed state effects that were independent of working memory load. These findings represent a step toward identifying how different brain regions exhibit varying vulnerability to the deleterious effects of SD on working memory.

Functional Specialization and Flexibility in Human Association Cortex

The association cortex supports cognitive functions enabling flexible behavior. Here, we explored the organization of human association cortex by mathematically formalizing the notion that a behavioral task engages multiple cognitive components, which are in turn supported by multiple overlapping brain regions. Application of the model to a large data set of neuroimaging experiments (N = 10 449) identified complex zones of frontal and parietal regions that ranged from being highly specialized to highly flexible. The network organization of the specialized and flexible regions was explored with an independent resting-state fMRI data set (N = 1000). Cortical regions specialized for the same components were strongly coupled, suggesting that components function as partially isolated networks. Functionally flexible regions participated in multiple components to different degrees. This heterogeneous selectivity was predicted by the connectivity between flexible and specialized regions. Functionally flexible regions might support binding or integrating specialized brain networks that, in turn, contribute to the ability to execute multiple and varied tasks.

Skull stripping using graph cuts.

Removal of non-brain tissues, particularly dura, is an important step in enabling accurate measurement of brain structures. Many popular methods rely on iterative surface deformation to fit the brain boundary and tend to leave residual dura. Similar to other approaches, the method proposed here uses intensity thresholding followed by removal of narrow connections to obtain a brain mask. However, instead of using morphological operations to remove narrow connections, a graph theoretic image segmentation technique was used to position cuts that isolate and remove dura. This approach performed well on both the standardized IBSR test data sets and empirically derived data. Compared to the Hybrid Watershed Algorithm (HWA; (Segonne et al., 2004)) the novel approach achieved an additional 10-30% of dura removal without incurring further brain tissue erosion. The proposed method is best used in conjunction with HWA as the errors produced by the two approaches often occur at different locations and cancel out when their masks are combined. Our experiments indicate that this combination can substantially decrease and often fully avoid cortical surface overestimation in subsequent segmentation.

Estimates of Segregation and Overlap of Functional Connectivity Networks in the Human Cerebral Cortex

The organization of the human cerebral cortex has recently been explored using techniques for parcellating the cortex into distinct functionally coupled networks. The divergent and convergent nature of cortico-cortical anatomic connections suggests the need to consider the possibility of regions belonging to multiple networks and hierarchies among networks. Here we applied the Latent Dirichlet Allocation (LDA) model and spatial independent component analysis (ICA) to solve for functionally coupled cerebral networks without assuming that cortical regions belong to a single network. Data analyzed included 1000 subjects from the Brain Genomics Superstruct Project (GSP) and 12 high quality individual subjects from the Human Connectome Project (HCP). The organization of the cerebral cortex was similar regardless of whether a winner-take-all approach or the more relaxed constraints of LDA (or ICA) were imposed. This suggests that large-scale networks may function as partially isolated modules. Several notable interactions among networks were uncovered by the LDA analysis. Many association regions belong to at least two networks, while somatomotor and early visual cortices are especially isolated. As examples of interaction, the precuneus, lateral temporal cortex, medial prefrontal cortex and posterior parietal cortex participate in multiple paralimbic networks that together comprise subsystems of the default network. In addition, regions at or near the frontal eye field and human lateral intraparietal area homologue participate in multiple hierarchically organized networks. These observations were replicated in both datasets and could be detected (and replicated) in individual subjects from the HCP.

Hippocampal region-specific contributions to memory performance in normal elderly.

To investigate the relationship between regional hippocampal volume and memory in healthy elderly, 147 community-based volunteers, aged 55-83years, were evaluated using magnetic resonance imaging, the Groton Maze Learning Test, Visual Reproduction and the Rey Auditory Verbal Learning Test. Hippocampal volumes were determined by interactive volumetry. We found greater age-related reduction in the volume of the hippocampal head relative to the tail. Right hippocampal tail volume correlated with spatial memory on the Groton Maze Learning Test while left hippocampal body volume was associated with delayed verbal memory. These associations were independent of age, sex, education and speed of processing and support the notion of functional differentiation along the long axis of the hippocampus.

Lapsing when sleep deprived: Neural activation characteristics of resistant and vulnerable individuals

Lapses of attention, in the form of delayed responses to salient stimuli, increase in frequency for some but not all persons after sleep deprivation (SD). To identify patterns of task-related brain activation that might explain differences in vulnerability to SD, we performed fMRI on participants during a visual, selective attention task. We analyzed the correct responses in a trial-by-trial fashion to model the effects of response time. Stimulus contrast was varied to modulate perceptual difficulty. Attentional lapses and low-contrast stimuli were independently associated with increased signal in fronto-parietal regions associated with biasing attention. Sleep-deprived vulnerable individuals showed reduced top down fronto-parietal signal across all levels of image contrast and this reduction was particularly significant during lapses. There was concurrent reduction in extrastriate cortex and thalamus activation. Non-vulnerable persons showed a trend towards higher top-down biasing of attention and preserved visual cortex activation during SD lapses. A major contributor to performance degradation in SD appears to be a reduction in top-down biasing of attention that is independent of task difficulty.

Improvement of brain segmentation accuracy by optimizing non-uniformity correction using N3

Smoothly varying and multiplicative intensity variations within MR images that are artifactual, can reduce the accuracy of automated brain segmentation. Fortunately, these can be corrected. Among existing correction approaches, the nonparametric non-uniformity intensity normalization method N3 (Sled, J.G., Zijdenbos, A.P., Evans, A.C., 1998. Nonparametric method for automatic correction of intensity nonuniformity in MRI data. IEEE Trans. Med. Imag. 17, 87-97.) is one of the most frequently used. However, at least one recent study (Boyes, R.G., Gunter, J.L., Frost, C., Janke, A.L., Yeatman, T., Hill, D.L.G., Bernstein, M.A., Thompson, P.M., Weiner, M.W., Schuff, N., Alexander, G.E., Killiany, R.J., DeCarli, C., Jack, C.R., Fox, N.C., 2008. Intensity non-uniformity correction using N3 on 3-T scanners with multichannel phased array coils. NeuroImage 39, 1752-1762.) suggests that its performance on 3 T scanners with multichannel phased-array receiver coils can be improved by optimizing a parameter that controls the smoothness of the estimated bias field. The present study not only confirms this finding, but additionally demonstrates the benefit of reducing the relevant parameter values to 30-50 mm (default value is 200 mm), on white matter surface estimation as well as the measurement of cortical and subcortical structures using FreeSurfer (Martinos Imaging Centre, Boston, MA). This finding can help enhance precision in studies where estimation of cerebral cortex thickness is critical for making inferences.

Cortical surface-based searchlight decoding.

Local voxel patterns of fMRI signals contain specific information about cognitive processes ranging from basic sensory processing to high level decision making. These patterns can be detected using multivariate pattern classification, and localization of these patterns can be achieved with searchlight methods in which the information content of spherical sub-volumes of the fMRI signal is assessed. The only assumption made by this approach is that the patterns are spatially local. We present a cortical surface-based searchlight approach to pattern localization. Voxels are grouped according to distance along the cortical surface-the intrinsic metric of cortical anatomy-rather than Euclidean distance as in volumetric searchlights. Using a paradigm in which the category of visually presented objects is decoded, we compare the surface-based method to a standard volumetric searchlight technique. Group analyses of accuracy maps produced by both methods show similar distributions of informative regions. The surface-based method achieves a finer spatial specificity with comparable peak values of significance, while the volumetric method appears to be more sensitive to small informative regions and might also capture information not located directly within the gray matter. Furthermore, our findings show that a surface centered in the middle of the gray matter contains more information than to the white-gray boundary or the pial surface.

Functional connectivity during rested wakefulness predicts vulnerability to sleep deprivation

Significant inter-individual differences in vigilance decline following sleep deprivation exist. We characterized functional connectivity in 68 healthy young adult participants in rested wakefulness and following a night of total sleep deprivation. After whole brain signal regression, functionally connected cortical networks during the well-rested state exhibited reduced correlation following sleep deprivation, suggesting that highly integrated brain regions become less integrated during sleep deprivation. In contrast, anti-correlations in the well-rested state became less so following sleep deprivation, suggesting that highly segregated networks become less segregated during sleep deprivation. Subjects more resilient to vigilance decline following sleep deprivation showed stronger anti-correlations among several networks. The weaker anti-correlations overlapped with connectivity alterations following sleep deprivation. Resilient individuals thus evidence clearer separation of highly segregated cortical networks in the well-rested state. In contrast to corticocortical connectivity, subcortical-cortical connectivity was comparable across resilient and vulnerable groups despite prominent state-related changes in both groups. Because sleep deprivation results in a significant elevation of whole brain signal amplitude, the aforesaid signal changes and group contrasts may be masked in analyses omitting their regression, suggesting possible value in regressing whole brain signal in certain experimental contexts.