Comparison of minimally invasive to standard temporal lobectomy approaches to epilepsy surgery: Seizure relief and visual confrontation naming outcomes.

The purpose of this study was to systematically examine three different surgical approaches in treating left medial temporal lobe epilepsy (mTLE) (viz., subtemporal selective amygdalohippocampectomy [subSAH], stereotactic laser amygdalohippocampotomy [SLAH], and anterior temporal lobectomy [ATL]), to determine which procedures are most favorable in terms of visual confrontation naming and seizure relief outcome. This was a retrospective study of 33 adults with intractable mTLE who underwent left temporal lobe surgery at three different epilepsy surgery centers who also underwent pre-, and at least 6-month post-surgical neuropsychological testing. Measures included the Boston Naming Test (BNT) and the Engel Epilepsy Surgery Outcome Scale. Fisher's exact tests revealed a statistically significant decline in naming in ATLs compared to SLAHs, but no other significant group differences. 82% of ATL and 36% of subSAH patients showed a significant naming decline whereas no SLAH patient (0%) had a significant naming decline. Significant postoperative naming improvement was seen in 36% of SLAH patients in contrast to 9% improvement in subSAH patients and 0% improvement in ATLs. Finally, there were no statistically significant differences between surgical approaches with regard to seizure freedom outcome, although there was a trend towards better seizure relief outcome among the ATL patients. Results support a possible benefit of SLAH in preserving visual confrontation naming after left TLE surgery. While result interpretation is limited by the small sample size, findings suggest outcome is likely to differ by surgical approach, and that further research on cognitive and seizure freedom outcomes is needed to inform patients and providers of potential risks and benefits with each.

Intracranial investigation of piriform cortex epilepsy during odor presentation.

The piriform cortex (PC) is part of the olfactory system, principally receiving input from the lateral olfactory tract and projecting to downstream components of the olfactory network, including the amygdala. Based on preclinical studies, PC is vulnerable to injury and can be easily kindled as an onset site for seizures. While the role of PC in human epilepsy has been studied indirectly and the subject of speculation, cases of demonstrated PC seizure onset from direct intracranial recording are rare. We present a pediatric patient with drug-resistant focal reflex epilepsy and right mesial temporal sclerosis with habitual seizures triggered by coconut aroma. The patient underwent stereoelectroencephalography with implantation of olfactory cortices including PC, through which we identified PC seizure onset, mapped high-frequency activity associated with presentation of olfactory stimuli and performance on cognitive tasks, and reproduced habitual seizures via cortical stimulation of PC. Coconut odor did not trigger seizures in our work with the patient. Surgical workup resulted in resection of the patient's right amygdala, PC, and mesial temporal pole, following which she has been seizure free for 20 months without functional decline in cognition or smell. Histological findings from resected tissue showed astrogliosis and subpial gliosis.

A virtual reality paradigm with dynamic scene stimuli for use in memory research.

Episodic memory may essentially be memory for one's place within a temporally unfolding scene from a first-person perspective. Given this, pervasively used static stimuli may only capture one small part of episodic memory. A promising approach for advancing the study of episodic memory is immersing participants within varying scenes from a first-person perspective. We present a pool of distinct scene stimuli for use in virtual environments and a paradigm that is implementable across varying levels of immersion on multiple virtual reality (VR) platforms and adaptable to studying various aspects of scene and episodic memory. In our task, participants are placed within a series of virtual environments from a first-person perspective and guided through a virtual tour of scenes during a study phase and a test phase. In the test phase, some scenes share a spatial layout with studied scenes; others are completely novel. In three experiments with varying degrees of immersion, we measure scene recall, scene familiarity-detection during recall failure, the subjective experience of déjà vu, the ability to predict the next turn on a tour, the subjective sense of being able to predict the next turn on a tour, and the factors that influence memory search and the inclination to generate candidate recollective information. The level of first-person immersion mattered to multiple facets of episodic memory. The paradigm presents a useful means of advancing mechanistic understanding of how memory operates in realistic dynamic scene environments, including in combination with cognitive neuroscience methods such as functional magnetic resonance imaging and electrophysiology.

Opening a conceptual space for metamemory experience.

The experiences associated with remembering, including metamemory feelings about the act of remembering and attempts at remembering, are not often integrated into general accounts of memory. For example, David Rubin (2022) proposes a unified, three-dimensional conceptual space for mapping memory states, a map that does not systematically specify metamemory feelings. Drawing on Rubin's model, we define a distinct role for metamemory in relation to first-order memory content. We propose a fourth dimension for the model and support the proposal with conceptual, neurocognitive, and clinical lines of reasoning. We use the modified model to illustrate several cases, and show how it helps to conceptualize a new category of memory state: autonoetic knowing, exemplified by déjà vu. We also caution not to assume that memory experience is directly correlated with or caused by memory content, an assumption Tulving (1989) labeled the doctrine of concordance.

Subjective distinguishability of seizure and non-seizure Déjà Vu: A case report, brief literature review, and research prospects.

Roughly two-thirds of all people report having experienced déjà vu-the odd feeling that a current experience is both novel and a repeat or replay of a previous, unrecalled experience. Reports of an association between déjà vu and seizure aura symptomatology have accumulated for over a century, and frequent déjà vu is also now known to be associated with focal seizures, particularly those of a medial temporal lobe (MTL) origin. A longstanding question is whether seizure-related déjà vu has the same basis and is the same subjective experience as non-seizure déjà vu. Survey research suggests that people who experience both seizure-related and non-seizure déjà vu can often subjectively distinguish between the two. We present a case of a person with a history of focal MTL seizures who reports having experienced both seizure-related and non-seizure common déjà vu, though the non-seizure type was more frequent during this person's youth than it is currently. The patient was studied with a virtual tour paradigm that has previously been shown to elicit déjà vu among non-clinical, young adult participants. The patient reported experiencing déjà vu of the common non-seizure type during the virtual tour paradigm, without associated abnormalities of the intracranial EEG. We situate this work in the context of broader ongoing projects examining the subjective correlates of seizures. The importance for memory research of virtual scenes, spatial tasks, virtual reality (VR), and this paradigm for isolating familiarity in the context of recall failure are discussed.

Stereotactic laser amygdalohippocampotomy for mesial temporal lobe epilepsy.

OBJECTIVE: To evaluate the outcomes 1 year and longer following stereotactic laser amygdalohippocampotomy for mesial temporal lobe epilepsy in a large series of patients treated over a 5-year period since introduction of this novel technique. METHODS: Surgical outcomes of a consecutive series of 58 patients with mesial temporal lobe epilepsy who underwent the surgery at our institution with at least 12 months of follow-up were retrospectively evaluated. A subgroup analysis was performed comparing patients with and without mesial temporal sclerosis. RESULTS: One year following stereotactic laser amygdalohippocampotomy, 53.4% (95% confidence interval [CI] = 40.8-65.7%) of all patients were free of disabling seizures (Engel I). Three of 9 patients became seizure-free following repeat ablation. Subgroup analysis showed that 60.5% (95% CI = 45.6-73.7%) of patients with mesial temporal sclerosis were free of disabling seizures as compared to 33.3% (95% CI = 15.0-58.5%) of patients without mesial temporal sclerosis. Quality of Life in Epilepsy-31 scores significantly improved at the group level, few procedure-related complications were observed, and verbal memory outcome was better than historical open resection data. INTERPRETATION: In an unselected consecutive series of patients, stereotactic laser amygdalohippocampotomy yielded seizure-free rates for patients with mesial temporal lobe epilepsy lower than, but comparable to, the outcomes typically associated with open temporal lobe surgery. Analogous to results from open surgery, patients without mesial temporal sclerosis fared less well. This novel procedure is an effective minimally invasive alternative to resective surgery. In the minority of patients not free of disabling seizures, laser ablation presents no barrier to additional open surgery. Ann Neurol 2018;83:575-587.

Mechanisms and Risk Factors Contributing to Visual Field Deficits following Stereotactic Laser Amygdalohippocampotomy.

Selective laser amygdalohippocampotomy (SLAH) is a minimally invasive surgical treatment for medial temporal lobe epilepsy. Visual field deficits (VFDs) are a significant potential complication. The objective of this study was to determine the relationship between VFDs and potential mechanisms of injury to the optic radiations and lateral geniculate nucleus. We performed a retrospective cross-sectional analysis of 3 patients (5.2%) who developed persistent VFDs after SLAH within our larger series (n = 58), 15 healthy individuals and 10 SLAH patients without visual complications. Diffusion tractography was used to evaluate laser catheter penetration of the optic radiations. Using a complementary approach, we evaluated evidence for focal microstructural tissue damage within the optic radiations and lateral geniculate nucleus. Overablation and potential heat radiation were assessed by quantifying ablation and choroidal fissure CSF volumes as well as energy deposited during SLAH.SLAH treatment parameters did not distinguish VFD patients. Atypically high overlap between the laser catheter and optic radiations was found in 1/3 VFD patients and was accompanied by focal reductions in fractional anisotropy where the catheter entered the lateral occipital white matter. Surprisingly, lateral geniculate tissue diffusivity was abnormal following, but also preceding, SLAH in patients who subsequently developed a VFD (all p = 0.005).In our series, vision-related complications following SLAH, which appear to occur less frequently than following open temporal lobe -surgery, were not directly explained by SLAH treatment parameters. Instead, our data suggest that variations in lateral geniculate structure may influence susceptibility to indirect heat injury from transoccipital SLAH.

Trends in the ambulatory management of headache: analysis of NAMCS and NHAMCS data 1999-2010.

BACKGROUND: Headache is a frequent complaint and among the most common reasons for visiting a physician. OBJECTIVE: To characterize trends from 1999 through 2010 in the management of headache. DESIGN: Longitudinal trends analysis. DATA: Nationally representative sample of visits to clinicians for headache from the National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey, excluding visits with "red flags," such as neurologic deficit, cancer, or trauma. MAIN MEASURES: Use of advanced imaging (CT/MRI), opioids/barbiturates, and referrals to other physicians (guideline-discordant indicators), as well as counseling on lifestyle modifications and use of preventive medications including verapamil, topiramate, amitriptyline, or propranolol (guideline-concordant during study period). We analyzed results using logistic regression, adjusting for patient and clinician characteristics, and weighted to reflect U.S. population estimates. Additionally, we stratified findings based on migraine versus non-migraine, acute versus chronic symptoms, and whether the clinician self-identified as the primary care physician. KEY RESULTS: We identified 9,362 visits for headache, representing an estimated 144 million visits during the study period. Nearly three-quarters of patients were female, and the mean age was approximately 46 years. Use of CT/MRI rose from 6.7% of visits in 1999-2000 to 13.9% in 2009-2010 (unadjusted p < 0.001), and referrals to other physicians increased from 6.9 % to 13.2% (p = 0.005). In contrast, clinician counseling declined from 23.5 % to 18.5% (p = 0.041). Use of preventive medications increased from 8.5 % to 15.9% (p = 0.001), while opioids/barbiturates remained unchanged, at approximately 18%. Adjusted trends were similar, as were results after stratifying by migraine versus non-migraine and acute versus chronic presentation. Primary care clinicians had lower odds of ordering CT/MRI (OR 0.56 [0.42, 0.74]). CONCLUSIONS: Contrary to numerous guidelines, clinicians are increasingly ordering advanced imaging and referring to other physicians, and less frequently offering lifestyle counseling to their patients. The management of headache represents an important opportunity to improve the value of U.S. healthcare.

Glutamatergic signaling from the parabrachial nucleus plays a critical role in hypercapnic arousal.

The mechanisms of arousal from apneas during sleep in patients suffering from obstructive sleep apnea are not well understood. However, we know that respiratory chemosensory pathways converge on the parabrachial nucleus (PB), which sends glutamatergic projections to a variety of forebrain structures critical to arousal, including the basal forebrain, lateral hypothalamus, midline thalamus, and cerebral cortex. We tested the role of glutamatergic signaling in this pathway by developing an animal model for repetitive CO2 arousals (RCAs) and investigating the effect of deleting the gene for the vesicular glutamate transporter 2 (Vglut2) from neurons in the PB. We used mice with lox P sequences flanking exon2 of the Vglut2 gene, in which adeno-associated viral vectors containing genes encoding Cre recombinase and green fluorescent protein were microinjected into the PB to permanently and selectively disrupt Vglut2 expression while labeling the affected neurons. We recorded sleep in these mice and then investigated the arousals during RCA. Vglut2 deletions that included the external lateral and lateral crescent subdivisions of the lateral PB more than doubled the latency to arousal and resulted in failure to arouse by 30 s in >30% of trials. By contrast, deletions that involved the medial PB subdivision had minimal effects on arousal during hypercapnia but instead increased non-rapid eye movement (NREM) sleep by ∼43% during the dark period, and increased delta power in the EEG during NREM sleep by ∼50%. Our results suggest that glutamatergic neurons in the lateral PB are necessary for arousals from sleep in response to CO2, while medial PB glutamatergic neurons play an important role in promoting spontaneous waking.

Bayesian Estimation Improves Prediction of Outcomes after Epilepsy Surgery.

Low power is a problem in many fields, as underpowered studies that find a statistically significant result will exaggerate the magnitude of the observed effect size. We quantified the statistical power and magnitude error of studies of epilepsy surgery outcomes. The median power across all studies was 14%. Studies with a median sample size or less (n<=56) and a statistically significant result exaggerated the true effect size by a factor of 5.4 (median odds ratio 9.3 vs. median true odds ratio 1.7), while the Bayesian estimate of the odds ratio only exaggerated the true effect size by a factor of 1.6 (2.7 vs. 1.7). We conclude that Bayesian estimation of odds ratio attenuates the exaggeration of significant effect sizes in underpowered studies. This approach could help improve patient counseling about the chance of seizure freedom after epilepsy surgery.

The involvement of the piriform cortex in non-lesional temporal lobe epilepsy: an uncommon component of the epileptogenic network.

The piriform cortex is recognized as highly epileptogenic in rodents, yet its electrophysiological role in human epilepsy remains understudied. Recent surgical outcomes have suggested potential benefits in resecting the piriform cortex for cases of medial temporal lobe epilepsy. However, little is known about its electrophysiological activity in human epilepsy. This case-series study aimed to explore the electrophysiological role of the piriform cortex within the epileptogenic network among patients with suspected temporal lobe epilepsy. Participants were recruited from Emory University Hospital or Children's Healthcare of Atlanta, with non-lesional frontotemporal or temporal lobe hypotheses, undergoing stereoelectroencephalographic studies. Specifically, focus was placed on patients with one or more electrode contacts in the piriform cortex. Primary objectives included determining piriform cortex involvement within the electrophysiologically defined epileptogenic network and assessing the effects of electrical stimulation. Twenty-two patients were included in the study. Notably, only one patient exhibited piriform cortex involvement at seizure onset, associated with an olfactory aura. Two patients showed early piriform cortex involvement, while others displayed late or no involvement. Electrical stimulation of the piriform cortex induced after-discharges in three patients and replicated a habitual seizure in one. These findings present a contrast to surgical outcome studies, suggesting that the piriform cortex may not typically play a significant role in the epileptogenic network among patients with non-lesional temporal lobe epilepsy.

Selective Posterior Cerebral Artery Wada Better Predicts Good Memory and Naming Outcomes Following Selective Stereotactic Thermal Ablation for Medial Temporal Lobe Epilepsy Than Internal Carotid Artery Wada.

The conventional intracarotid amobarbital (Wada) test has been used to assess memory function in patients being considered for temporal lobe epilepsy (TLE) surgery. Minimally invasive approaches that target the medial temporal lobe (MTL) and spare neocortex are increasingly used, but a knowledge gap remains in how to assess memory and language risk from these procedures. We retrospectively compared results of two versions of the Wada test, the intracarotid artery (ICA-Wada) and posterior cerebral artery (PCA-Wada) approaches, with respect to predicting subsequent memory and language outcomes, particularly after stereotactic laser amygdalohippocampotomy (SLAH). We included all patients being considered for SLAH who underwent both ICA-Wada and PCA-Wada at a single institution. Memory and confrontation naming assessments were conducted using standardized neuropsychological tests to assess pre- to post-surgical changes in cognitive performance. Of 13 patients who initially failed the ICA-Wada, only one patient subsequently failed the PCA-Wada (p=0.003, two-sided binomial test with p 0 =0.5) demonstrating that these tests assess different brain regions or networks. PCA-Wada had a high negative predictive value for the safety of SLAH, compared to ICA-Wada, as none of the patients who underwent SLAH after passing the PCA-Wada experienced catastrophic memory decline (0 of 9 subjects, p <.004, two-sided binomial test with p 0 =0.5), and all experienced a good cognitive outcome. In contrast, the single patient who received a left anterior temporal lobectomy after failed ICA- and passed PCA-Wada experienced a persistent, near catastrophic memory decline. On confrontation naming, few patients exhibited disturbance during the PCA-Wada. Following surgery, SLAH patients showed no naming decline, while open resection patients, whose surgeries all included ipsilateral temporal lobe neocortex, experienced significant naming difficulties (Fisher's exact test, p <.05). These findings demonstrate that (1) failing the ICA-Wada falsely predicts memory decline following SLAH, (2) PCA-Wada better predicts good memory outcomes of SLAH for MTLE, and (3) the MTL brain structures affected by both PCA-Wada and SLAH are not directly involved in language processing.

Classifier for the Rapid Simultaneous Determination of Sleep-Wake States and Seizures in Mice.

Independent automated scoring of sleep-wake and seizures have recently been achieved; however, the combined scoring of both states has yet to be reported. Mouse models of epilepsy typically demonstrate an abnormal electroencephalographic (EEG) background with significant variability between mice, making combined scoring a more difficult classification problem for manual and automated scoring. Given the extensive EEG variability between epileptic mice, large group sizes are needed for most studies. As large datasets are unwieldy and impractical to score manually, automatic seizure and sleep-wake classification are warranted. To this end, we developed an accurate automated classifier of sleep-wake states, seizures, and the post-ictal state. Our benchmark was a classification accuracy at or above the 93% level of human inter-rater agreement. Given the failure of parametric scoring in the setting of altered baseline EEGs, we adopted a machine-learning approach. We created several multi-layer neural network architectures that were trained on human-scored training data from an extensive repository of continuous recordings of electrocorticogram (ECoG), left and right hippocampal local field potential (HPC-L and HPC-R), and electromyogram (EMG) in the murine intra-amygdala kainic acid model of medial temporal lobe epilepsy. We then compared different network models, finding a bidirectional long short-term memory (BiLSTM) design to show the best performance with validation and test portions of the dataset. The SWISC (sleep-wake and the ictal state classifier) achieved >93% scoring accuracy in all categories for epileptic and non-epileptic mice. Classification performance was principally dependent on hippocampal signals and performed well without EMG. Additionally, performance is within desirable limits for recording montages featuring only ECoG channels, expanding its potential scope. This accurate classifier will allow for rapid combined sleep-wake and seizure scoring in mouse models of epilepsy and other neurologic diseases with varying EEG abnormalities, thereby facilitating rigorous experiments with larger numbers of mice.

Ordinal regression increases statistical power to predict epilepsy surgical outcomes.

Studies of epilepsy surgery outcomes are often small and thus underpowered to reach statistically valid conclusions. We hypothesized that ordinal logistic regression would have greater statistical power than binary logistic regression when analyzing epilepsy surgery outcomes. We reviewed 10 manuscripts included in a recent meta-analysis which found that mesial temporal sclerosis (MTS) predicted better surgical outcomes after a stereotactic laser amygdalohippocampectomy (SLAH). We extracted data from 239 patients from eight studies that reported four discrete Engel surgical outcomes after SLAH, stratified by the presence or absence of MTS. The rate of freedom from disabling seizures (Engel I) was 64.3% (110/171) for patients with MTS compared to 44.1% (30/68) without MTS. The statistical power to detect MTS as a predictor for better surgical outcome after a SLAH was 29% using ordinal regression, which was significantly more than the 13% power using binary logistic regression (paired t-test, P < .001). Only 120 patients are needed for this example to achieve 80% power to detect MTS as a predictor using ordinal regression, compared to 210 patients that are needed to achieve 80% power using binary logistic regression. Ordinal regression should be considered when analyzing ordinal outcomes (such as Engel surgical outcomes), especially for datasets with small sample sizes.

Ultrafast simulation of large-scale neocortical microcircuitry with biophysically realistic neurons.

Understanding the activity of the mammalian brain requires an integrative knowledge of circuits at distinct scales, ranging from ion channel gating to circuit connectomics. Computational models are regularly employed to understand how multiple parameters contribute synergistically to circuit behavior. However, traditional models of anatomically and biophysically realistic neurons are computationally demanding, especially when scaled to model local circuits. To overcome this limitation, we trained several artificial neural network (ANN) architectures to model the activity of realistic multicompartmental cortical neurons. We identified an ANN architecture that accurately predicted subthreshold activity and action potential firing. The ANN could correctly generalize to previously unobserved synaptic input, including in models containing nonlinear dendritic properties. When scaled, processing times were orders of magnitude faster compared with traditional approaches, allowing for rapid parameter-space mapping in a circuit model of Rett syndrome. Thus, we present a novel ANN approach allowing for rapid, detailed network experiments using inexpensive and commonly available computational resources.

The Referential Montage Inadequately Localizes Corticocortical Evoked Potentials in Stereoelectroencephalography.

PURPOSE: Corticocortical evoked potentials (CCEPs) resulting from single pulse electrical stimulation are increasingly used to understand seizure networks, as well as normal brain connectivity. However, we observed that when using depth electrodes, traditional measures of CCEPs amplitude using a referential montage can be falsely localizing, often to white matter. METHODS: We pooled 27 linear electrode arrays targeting the amygdala, hippocampus, or cingulate cortex from eight participants. Using postoperative imaging, we classified contacts as being in gray matter, white matter, or bordering each and measured the amplitude using the root-mean-squared deviation from baseline in a referential, common average, bipolar, or Laplacian montage. RESULTS: Of 27 electrode contacts, 25 (93%) had a significantly higher mean amplitude when in gray matter than in white matter using a Laplacian montage, which was significantly more than the 12 of 27 electrodes (44%) when using a referential montage ( P = 0.0003, Fisher exact test). The area under the curve for a receiver operating characteristic classifying contacts as gray or white matter was significantly higher for either the Laplacian (0.79) or the bipolar (0.72) montage when compared with either the common average (0.56) or the referential (0.51) montage ( P ≤ 0.005, bootstrap). CONCLUSIONS: Both the Laplacian and bipolar montages were superior to the common average or referential montage in localizing CCEPs to gray matter. These montages may be more appropriate for interpreting CCEPs when using depth electrodes than the referential montage, which has typically been used in prior studies of CCEPs with subdural grids.

Loss of efferent projections of the hippocampal formation in the mouse intrahippocampal kainic acid model.

Unilateral intrahippocampal injection of kainic acid is used as a model of medial temporal lobe epilepsy and provides a platform to study the mechanisms of epilepsy. Here, we used an AAV-9 EYFP-tagged viral vector as an anterograde tracer, injected into the dorsal and ventral hippocampus after kainic acid injection, to map out the efferent hippocampal projections after the development of spontaneous seizures in this model. The purpose of the study was to identify the extent of changes in hippocampal efferent system in several brain regions that receive significant inputs from the hippocampus. Loss of efferent hippocampal fibers was greatest in the retrosplenial cortex where neuronal loss was also observed. Loss of fibers was also observed in the fornix without any specific effect in the lateral mammillary nuclei. Although expected, these observations provide further evidence of the broader network effects as a result of hippocampal cell loss.

Electrical Wada for pre-surgical memory testing: a case report.

We report a case study of a surgical candidate, a 51-year-old woman with left temporal lobe epilepsy, who failed a left injection intracarotid amobarbital procedure (e.g., Wada test), scoring 0 of 8 items. This raised concerns for postoperative memory decline. However, the patient was uninterested in a neuromodulatory approach and wished to be reconsidered for surgery. A stereotactic laser amygdalohippocampotomy (SLAH) was considered, encouraging the need for an alternative test to evaluate risk of memory decline. We developed a novel approach to testing memory during stimulation of a depth electrode implanted in the hippocampus, i.e., an electric Wada. During multiple stimulation trials across a range of amplitudes, the patient scored up to 8 of 8 items, which suggested strong contralateral memory support. The surgical team proceeded with a radiofrequency ablation and a subsequent SLAH. The patient remains seizure-free at 12 months post SLAH with no evidence of verbal or visuospatial memory decline based on a post-surgical neuropsychological battery. We believe that this case study provides a proof of concept for the feasibility and possible utility of an electric version of the Wada procedure. Future studies are needed to develop an optimal paradigm and to validate this approach.

Radiological identification of temporal lobe epilepsy using artificial intelligence: a feasibility study.

Temporal lobe epilepsy is associated with MRI findings reflecting underlying mesial temporal sclerosis. Identifying these MRI features is critical for the diagnosis and management of temporal lobe epilepsy. To date, this process relies on visual assessment by highly trained human experts (e.g. neuroradiologists, epileptologists). Artificial intelligence is increasingly recognized as a promising aid in the radiological evaluation of neurological diseases, yet its applications in temporal lobe epilepsy have been limited. Here, we applied a convolutional neural network to assess the classification accuracy of temporal lobe epilepsy based on structural MRI. We demonstrate that convoluted neural networks can achieve high accuracy in the identification of unilateral temporal lobe epilepsy cases even when the MRI had been originally interpreted as normal by experts. We show that accuracy can be potentiated by employing smoothed grey matter maps and a direct acyclic graphs approach. We further discuss the foundations for the development of computer-aided tools to assist with the diagnosis of epilepsy.