Authors :
Presenting Author: Lucas Sainburg, BS – Vanderbilt University
Baxter Rogers, PhD – Department of Radiology and Radiological Sciences – Vanderbilt University Medical Center; Catie Chang, PhD – Department of Electrical and Computer Engineering – Vanderbilt University; Dario Englot, MD, PhD – Department of Neurological Surgery – Vanderbilt University Medical Center; Victoria Morgan, PhD – Department of Radiology and Radiological Sciences – Vanderbilt University Medical Center
Rationale:
Patients with temporal lobe epilepsy (TLE) often have transient abnormal electrophysiological activity in their epileptic hippocampus interictally. Prior work has used functional MRI (fMRI) to study the transient functional connectivity (FC) abnormalities of patients with TLE by assessing the dynamic FC of the hippocampus over minutes.1 Here we used a recently developed method2 to investigate very brief (~seconds) interictal FC dynamics of the hippocampus in patients with TLE.
Methods:
This study included 96 healthy controls (37.9 ± 13.4 yrs), split into training (n = 70) and testing (n = 26) groups, along with 23 patients with right TLE (RTLE) (42.5 ± 12.6 yrs). All participants had a T1-weighted scan (1x1x1 mm
3) and 20 minutes of resting-state fMRI (TR = 2 s, 3x3x4 mm
3) acquired on a 3T MRI scanner. T1-weighted images were used to segment the brain into 111 regions.
Normalized preprocessed fMRI timeseries from each pair of brain regions were multiplied at each timepoint to obtain the edge timeseries for each pair of regions2, representing the dynamic FC at each timepoint (every 2 s) (Fig 1A). The 110 edge timeseries to the right anterior hippocampus (seizure onset) were extracted and the FC at timepoints of high co-fluctuations across all 110 edges were identified as “co-fluctuation events”. The co-fluctuation events from the training controls were clustered into recurring spatial control co-fluctuation patterns of the anterior hippocampus. All co-fluctuation events from the patients and the testing controls were then classified as one of these control patterns (Figure 1B).
The distance between each participant’s co-fluctuation events and the associated control pattern was computed. The occurrence rate of each pattern was also computed for each participant. Distances and occurrence rates of each pattern were compared between patients and testing controls using Wilcoxon rank-sum tests.
Results:
Three right anterior hippocampus co-fluctuation patterns were detected in the training controls (Figure 2A). Pattern 1 had positive FC to medial temporal, subcortical, and default mode network regions along with negative FC to sensorimotor and lateral frontal regions. Pattern 2 had positive FC to temporal, occipital, and sensorimotor regions along with negative FC to the thalami. Pattern 3 had positive FC to medial temporal, insular, and medial frontal regions along with negative FC to occipital regions.
Patients with RTLE had a higher distance from Pattern 1 than testing controls (punc < 0.05, Figure 2B) and a higher occurrence rate of Pattern 3 than testing controls (punc < 0.05, Figure 2C).