Inter-hemispheric connectivity as a predictor of lesion outcomes

Abstract

Brain lesions have been used to localize cognitive processes, dating as far back to Broca’s discovery of speech-related regions. Even after strides in understanding the brain, predicting functional outcomes using the lesion’s location is difficult. Outcomes following damage to the brain lead to varying results among individuals. Lesions to similar regions can result in both debilitating or undetectable outcomes. Only recently has research shifted to understanding principles that dictate outcomes. Our studies explore inter-hemispheric connections to understand differences in long-term outcomes.

Analyses in Chapter 2 establish a metric for estimating inter-hemispheric connections over multiple domains. These metrics represent the white matter as a network using connection strengths. Hemispheric specialization is analyzed using metrics of inter-hemispheric connections and is compared to metrics of both intra- and inter-hemispheric connections (whole-brain). Our results confirm that inter-hemispheric connections reflect whether a function is isolated to a hemisphere or shared between hemispheres. Importantly, the whole-brain metrics fail to capture this aspect. Therefore, if an area is strongly connected to the mirroring region of the opposite hemisphere, both are likely to participate in the same functions.

Chapter 3 applies lesions on the basis of inter-hemispheric connections and then simulates the functional connectivity. Outcomes from such simulations can be compared to healthy networks, and the deficits are attributed to the topology of the damaged regions. Deficits from damage to regions containing the most inter-hemispheric connections are comparable to damage to most connected regions. Long-term outcomes are related to the capacity for rebuilding lost circuits to recover function. Our results find that regions with the most inter-hemispheric connections are also the most robust. Robustness is the ability to recruit redundant regions for compensation. Brain regions with many inter-hemispheric connections possess redundant paths through which other regions can replace lost functions. Connections supporting these properties are explored by altering symmetric connections that project to both mirrored regions. For example, a connection from Broca’s area to both dorsolateral PFC would be rewired so only the left or right connection remains. We find that these regions are no longer robust after rewiring. Damage to regions with many inter-hemispheric connections undergoes compensation through the opposite hemisphere.

In Chapter 4, inter-hemispheric metrics are evaluated against an estimate of long-term outcomes. We extracted fMRI coordinates from publications identifying the lesion location with the largest behavioral deficit. Extracted coordinate counts are used as an estimate of outcomes. We reasoned that the publication frequency for a lesion is dependent on the severity of behavioral outcomes. For example, the regions responsible for depression are understudied compared to regions responsible for paralysis. Overall, our results found that both sets of metrics are predictive of outcome, but inter-hemispheric metrics had significantly better performance.

Our analysis suggests that inter-hemispheric connections predict long-term outcomes due to the capacity to recruit redundant circuits. Therefore, brain regions like Broca’s area are more likely to result in behavioral deficits after unilateral damage compared to the dorsolateral PFC because Broca’s area has fewer inter-hemispheric connections whereas dorsolateral PFC has many.