Structural Subnetwork Associated with Specific Domains of Cognitive Function in TLE
Abstract number :
3.367
Submission category :
5. Neuro Imaging / 5A. Structural Imaging
Year :
2024
Submission ID :
125
Source :
www.aesnet.org
Presentation date :
12/9/2024 12:00:00 AM
Published date :
Authors :
Karla Batista Garcia-Ramo, PhD – Queen's University
Spencer Finn, MSc – Queen's University
Theodore Aliyianis, MSc – Queen's University
Adam Falah, BSc – Queen's University
Brooke Beattie, BSc – Queen's University
Donald Brien, MSc – Queen's University
Garima Shukla, MBBS, MD, DM, FRCPC – Queen's University
Lysa Boisse Lomax, MD, MSc, FRCPC, CSCN (EEG) – Queen's University
Stephen Scott, PhD – Queen's University
Jason Gallivan, BA, MSc, PhD – Queen's University
Presenting Author: Gavin Winston, BM BCh, MA, PhD, EMBA, FRCP, CSCN (EEG) – Queen's University
Rationale: Epilepsy is a network disorder and the disruption of neural networks can adversely affect neuropsychological function. Even though previous studies have related cognitive dysfunction to properties of neural networks in some types of epilepsy, a clear association between the two has not been delineated. The feasibility of the Kinarm robotic system to evaluate various sensorimotor and cognitive functions in individuals with epilepsy has been demonstrated. The present study aims to relate cognitive and sensorimotor dysfunction detected by robotic assessment to altered structural connectivity in participants with temporal lobe epilepsy (TLE).
Methods: A sample of 33 patients with TLE and 24 controls were included. All participants performed twelve tasks using robotic technology (Kinarm) to evaluate motor, cognitive, and sensory domains. A principal component analysis (PCA) of the performance on these tasks was conducted. Structural connectivity matrices were obtained from multi-shell diffusion weighted imaging using the Micapipe pipeline (https://github.com/MICA-MNI/micapipe) for processing. Statistical analysis was conducted using a network-based statistical approach combined with the features of machine learning.
Results: Patients showed significantly poorer performance in nine of the twelve tests evaluated, showing impairment across cognitive domains, including executive function, attention, memory, and psychomotor speed. The PCA yielded three orthogonal components accounting for 68% of the variance (Figure 1). A structural subnetwork of 38 nodes and 41 edges was significant associated only with the second component (Figure 2, R = 0.6, p-value < 0.01). This subnetwork includes connections between regions of the temporal lobe (superior temporal sulcus, amygdala, hippocampus, transverse temporal gyrus), occipitotemporal cortex, basal ganglia nucleus (thalamus and caudate), frontal lobe (frontal inferior, prefrontal and precentral gyrus), parietal lobe (intraparietal sulcus, parietal inferior), lateral fissure, insula and cerebellum. The circular sulcus of the insula was the node with highest degree. Notably, this second component contains tests that are more associated with executive function as well as tests of attention.
Neuro Imaging