Career Center AES Connect

Abstracts

Atypical Intrinsic Neural Timescales in Temporal Lobe Epilepsy

Abstract number : 2.186
Submission category : 5. Neuro Imaging / 5B. Functional Imaging
Year : 2022
Submission ID : 2204336
Source : www.aesnet.org
Presentation date : 12/4/2022 12:00:00 PM
Published date : Nov 22, 2022, 05:24 AM

Authors :
Ke Xie, Msc – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Jessica Royer, PsyD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Sara Lariviere, Msc – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Raul Rodriguez-Cruces, PhD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Reinder Vos de Wael, PhD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Bo-yong Park, PhD – Montreal Neurological Institute, McGill University; Department of Data Science, Inha University, Incheon, Republic of Korea; Center for Neuroscience Imaging Research, Institute for Basic Science, Suwon, Republic of Korea; Hans Auer, Bsc – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Shahin Tavakol, Msc – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Jordan DeKraker, PhD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Chifaou Abdallah, PhD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Lorenzo Caciagli, PhD – Department of Biomedical Engineering, University of Pennsylvania, Philadelphia, Pennsylvania 19104, USA; Andrea Bernasconi, MD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Neda Bernasconi, PhD, MD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Birgit Frauscher, PhD, MD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada; Luis Concha, PhD – Brain Connectivity Laboratory, Institute of Neurobiology, Universidad Nacional Autónoma de Mexico, Mexico; Boris Bernhardt, PhD – Montreal Neurological Institute, McGill University, Montreal, Quebec, Canada

Rationale: Temporal lobe epilepsy (TLE) is the most common drug-resistant epilepsy in adults. Increasing evidence indicates functional alterations in the whole-brain level in TLE. However, whether activity profiles of individual brain regions may also be altered is unclear. Here, we capitalize on intrinsic neural timescales (INT)1—a measure that reflects regional neural integration and follows hierarchical spatial gradients and investigate INT alterations in TLE relative to controls. We then examine their associations with structural changes, and evaluate the utility of INT in predicting clinical characteristics.

Methods: We studied 46 individuals with drug-resistant TLE (18 males; 32.2±8.6 y; 31 left-TLE) and 44 age- and sex-matched controls (20 males; 32.8±12.0 y) who were aggregated from two neuroimaging centers. Participants underwent 3T magnetic resonance imaging (MRI), including T1-weighted MRI, diffusion MRI, and resting-state functional MRI (rs-fMRI). INT values were computed from the autocorrelation function of rs-fMRI timeseries,2 and compared between groups using linear models that controlled for age, sex, and site. We also examined associations with structural alterations and explored the effects of age and clinical characteristics. Lastly, we assessed the utility of INT for patient/control classification and seizure focus lateralization with machine learning techniques.

Results: In controls, there were longer INT in prefrontal and parieto-occipital cortices as well as the thalamus, indexing stronger long-range temporal autocorrelations of rs-fMRI signal. Conversely, INT were shorter in sensorimotor and paralimbic cortices, and in the amygdala, nucleus accumbens and pallidum. Comparisons of TLE and controls revealed global INT reductions [neocortex/(subcortex+hippocampus), Cohen’s d = -0.53/-0.56, p < 0.05]. Vertex-wise comparisons revealed INT reductions in TLE in bilateral precuneus, central and occipital cortices, and ipsilateral temporal cortex (pFWE < 0.05), and in bilateral hippocampus and thalamus (pFDR < 0.05), indicating faster changing neural activity at rest. Findings were consistent across sites (Figure 1). Patterns of INT reductions in TLE remained relatively unchanged when controlling for grey matter morphological and white matter microstructural anomalies. INT aggravated with advanced age and duration in TLE. Finally, supervised machine learning achieved INT-based discrimination of patients from controls (balanced accuracy, 5-fold: 76±2.65%; cross-site, 72-83%) and lateralization of the seizure focus in TLE (96±2.10%; 95-97%) with high accuracies and generalization (Figure 2).
Neuro Imaging