Temporal lobe epilepsy (TLE) is characterized by debilitating and progressive cognitive impairment, but there is significant variability in the nature and severity of impairment across patients. Cognitive phenotyping is a promising approach for understanding the heterogeneity within TLE. This 3-paper dissertation aimed to identify the neural correlates associated with cognitive phenotypes, investigate methods for defining phenotypes, and examine epilepsy and health-related factors associated with each phenotype. Study 1 (Reyes et al., 2019) identified four distinct cognitive phenotypes in TLE (N = 70; 36.14 average age; 13.34 average education; 52% female) based on neuropsychological measures of memory and language. Each phenotype was associated with a unique pattern of white matter (WM) abnormalities. Patients with generalized impairment demonstrated widespread WM alterations, those with domain-specific impairment demonstrated regional WM alterations, and those with no impairment demonstrated WM patterns similar to controls. Study 2 (N = 407; 36.36 average age; 13.22 average education; 55% female; Reyes et al., 2020) compared phenotype classifications based on a neuropsychological approach (clinically-driven) versus cluster analysis (data-driven). Both approaches identified three unique cognitive phenotypes with strong agreement (kappa=.716); however, cluster analysis misclassified 12% of impaired patients as having normal cognition. These findings led to the question: could a more robust, person-centered, data-driven approach improve phenotyping? Study 3 (N = 1,178; 37.76 average age; 13.94 average education; 57% female; Reyes et al., under review) used latent profile analysis (LPA) to test several models of cognitive phenotyping and adjudicate the impact of missing data to identify the “best” taxonomy. LPA revealed that the three-class model was the optimal solution (entropy=.816) and the most robust to missing data with a 98.98% agreement with an imputed dataset (kappa=.983). Preliminary analyses revealed lower subcortical volumes in patients with generalized impairment and higher intracranial volumes in those with an intact profile. There was a differential association between hyperlipidemia and cognitive performance across phenotypes. These studies demonstrate unique cognitive phenotypes exist within TLE that are stable across investigations and approaches and are characterized by distinct neural signatures. Knowledge of these phenotypes could drive cognitive and neuroanatomical taxonomies in epilepsy and enhance individualized prediction of cognitive trajectories.

Temporal lobe epilepsy (TLE) is the most common form of focal epilepsy in adults. Anterior temporal lobectomy (ATL), i.e., the surgical removal of the anterior portion of the affected temporal lobe, is a common treatment for patients with focal, refractory TLE. ATLs require the severing of several key white matter tracts and may theoretically initiate a cascade of degenerative changes in frontotemporal networks. Nonetheless, structural factors that give rise to post-surgical changes in frontotemporal networks are rarely studied longitudinally. Indeed, ATL presents a prime experimental design opportunity where one can assess for post-surgical change and answer mechanistic questions of executive dysfunction in TLE.

In this dissertation, we built on our past observations that a particular frontotemporal white matter tract, the uncinate fasciculus (UF), is critical to executive functioning in TLE. We tested the hypothesis that the severing of the temporal segment of the UF during ATL creates a cascade of degenerative events in interconnected brain regions that may mediate executive function. We used a multi-modal imaging approach (i.e., diffusion tensor imaging, or DTI combined with magnetic resonance imaging, or MRI) to test whether patients with TLE who undergo ATL show diffusional changes in the frontal section of the UF consistent with Wallerian degeneration. We determined whether these changes correlate with performance changes in executive function tasks. Specifically, we quantified white matter microstructure in sections of the UF before and after surgery, and measured the association between the DTI changes with changes in patients’ executive abilities (i.e., task-switching, verbal fluency, inhibition).

Our results showed that UF is a tract with distinct microstructural damage in both the frontal and temporal sections even before ATL. We also found that UF sections both exhibit statistically significant diffusivity changes following ATL on the side ipsilateral to surgery, and that the changes observed for frontal UF are concordant with Wallerian degeneration. Surprisingly, we found that patients undergoing ATL showed significant improvement on one aspect of executive functioning (i.e., task-switching), with no post-ATL changes in verbal fluency or inhibition at the group level. However, we found no correlations between changes in frontal UF diffusivity and task-switching performance.

Objective: We aimed to determine whether a novel, white matter (WM)

network-based approach (i.e. structural connectomes; SC) outperforms

traditional tract-based measures and hippocampal volume (HCV) for discriminating

temporal lobe epilepsy (TLE) patients with memory-impairment from those with normal

memory.

Methods: T1-and diffusion-weighted imaging (DWI), clinical, and

neuropsychological data were available for 81 patients with TLE and 61 healthy

controls. Five medial temporal lobe WM tracts were extracted from the DWI and

fractional anisotropy (FA) was calculated for each tract. Left and right HCVs were

derived using FreeSurfer. SCs were derived by performing probabilistic tractography

and measuring the number of connections among cortical regions. SCs were fed into

XGBoost, a robust tree-based classifier. The model was trained on 48 patients

from UCSD and tested on 33 patients from UCSF.

Results: Both SCs (76% accuracy) and tracts (73% accuracy) were better

classifiers of verbal memory performance than HCV (66\% accuracy) or

clinical variables (61% accuracy). Within the WM-based models, the SC

performed similar to or better than the tract-based models, with slightly

higher accuracy and positive predictive value (PPV; 0.81 for SC, 0.77 for

tracts). The SC appeared to provide more specific information regarding

abnormal connectivity contributing to verbal memory performance. Multivariate

models seemed to be most robust classifiers, providing the highest sensitivity

(0.95 for Tract+HCV) and specificity (0.67 for Connectome+HCV).

Conclusion: SCs and tract-based measures appear to outperform HCVs and clinical

variables for predicting memory impairment in TLE. However, network-based

models combined with HCVs may provide the best prediction accuracy.

Temporal lobe epilepsy (TLE) is a debilitating neurological condition that is physiologically characterized by pathological changes to the hippocampus and proximal tissues. Diffusion tensor imaging (DTI) has repeatedly shown that patients with TLE also have microstructural changes to temporal and extratemporal white matter unseen in traditional structural imaging. Previous DTI studies have shown decreased fractional anisotropy (FA), a measure representing directionality of diffusion, in patients with TLE compared to controls. The systematic decline of FA observed along major white matter tracts in TLE is often interpreted as a loss of integrity of white matter. However, FA is a nonspecific measure that is heavily influenced by not only axonal loss, but also by extracellular changes (i.e., edema, inflammation) and crossing fibers.

In 21 patients with TLE and 11 age-matched controls, we aim to better characterize underlying white matter changes in TLE using a new diffusion model, Restricted Spectrum Imaging (RSI) that separates intracellular changes in diffusion (i.e., neurite density; ND) from extracellular changes (isotropic free water; IF) and crossing fibers (CF). Using both voxelwise and tract-based analysis, we demonstrate that both FA and ND measures have a high level of spatial correspondence, meaning the loss of directionality is primarily driven by the loss of diffusion derived from restricted neurite components, rather than increases in free water or the effects of CF. In addition, changes in ND, but not FA, were associated with disease duration, suggesting that ND may provide a more specific marker of white matter disease burden in refractory TLE.

Temporal lobe epilepsy (TLE) is the most common form of focal epilepsy in adults. Many of those affected do not respond well to anti-seizure medications and require neurosurgical intervention. For these patients, locating the seizure focus to the left or right hemisphere is a critical first step in surgical planning. Through the use of Convolutional Neural Networks (CNN) that can survey medical diagnostic images in finer detail and extract more convoluted information than humans, we proposed that it will aid in the localization of epileptogenic regions (i.e., seizure generating regions) in the brain. We explored different pre-trained and novel CNN architectures trained on 364 patients with TLE from seven different epilepsy centers to ensure generalizability. Our findings show that a CNN model outperforms a more standard linear model like a Logistic Regression (LR) in performing the task of side-of-onset classification, with the best CNN model outperforming the average LR model by 17.39% in classification accuracy. The model also shows that information important for determining side-of-onset is not limited to the mesial temporal lobe (MTL) but is also located in extra-temporal regions like the parietal lobe, precentral gyrus, and cingulate gyrus. This study shows that the classification problem of Left versus Right TLE patients benefits from a more complex, nonlinear model and whole-brain information and therefore medical examination of TLE would benefit from incorporating machine learning to aid in clinician-led localization of the epileptogenic zone.

Objective

Methods

Results

Discussion