Senior Research Assistant
Summer B. Frandsen, Jordan H. Grafman, Shan H. Siddiqi
Summer B. Frandsen
Research Category: Neurosciences
Intro: Therapeutic brain stimulation can treat neuropsychiatric disorders by activating specific brain regions. Specific symptoms can be modulated by targeting the brain networks which, when damaged, cause said symptoms. Here, we studied patients with penetrating brain injury to identify symptom clusters modulated by damage to different brain circuits.
Methods: Questionnaires, including 184 neuropsychiatric symptoms, traits, and diagnoses, were administered to 196 participants with focal brain lesions. Using a normative database of brain connectivity data (n=1000), we estimated the functional connectivity profile of each participant’s lesion location. We computed whole-brain correlation between lesion connectivity and symptom severity, yielding “circuit maps” indicating the connectivity of lesions which modify given symptoms. We used k-means clustering to identify symptom clusters which mapped to common brain networks, quantified clustering strength, and tested for significance using permutation testing.
Results: We identified at least two distinct behavioral clusters (p=.034). Lesions modifying traits such as “carefree” and “activeness” connected to a circuit that includes the ventral tegmental area. Lesions modifying symptoms such as sadness and helplessness connected to a circuit that includes the dorsolateral prefrontal cortex.
Discussion: Distinct patterns of brain damage can modify distinct symptom clusters, which may inform targeting of therapeutic brain stimulation based on symptoms.
Introduction: Neuromodulation is a treatment where the brain is stimulated to improve disorders. Most current treatments target whole diseases such as depression or OCD. However, people frequently have multiple disorders with overlapping symptoms. To treat multiple disorders simultaneously, we aimed to group symptoms and target all these symptoms at the same time. We ran multiple analyses to find “treatment sites” for groups of symptom clusters rather than one disorder.
Methods: The dataset we used (n=196) has data with lesions and symptoms. We ran analyses to calculate how the lesions and symptoms were connected to certain parts of the brain. We then ran multiple clustering algorithms to group the data based on how similar/different these connections are from one another.
Results: When we added lesion data, we saw significant results for clustering not previously present with solely behavioral data. We identified at least two symptom clusters. The two-cluster option had a positive emotion cluster and a negative emotion cluster.
Discussion: Lesion data clearly improved our ability to cluster. In the future, we hope to run these analyses to better define groups of clusters for neuromodulation treatments.