Current Literature In Clinical Science
Planning Resective Surgery Using Structural Connectivity Modeling: The Next-Generation Presurgical Evaluation
Predicting Surgery Targets in Temporal Lobe Epilepsy through Structural Connectome Based Simulations. Hutchings F, Han CE, Keller SS, Weber B, Taylor PN, Kaiser M. PLoS Comput Biol 2015;11:e1004642.
Temporal lobe epilepsy (TLE) is a prevalent neurological disorder resulting in disruptive seizures. In the case of drug resistant epilepsy resective surgery is often considered. This is a procedure hampered by unpredictable success rates, with many patients continuing to have seizures even after surgery. In this study we apply a computational model of epilepsy to patient specific structural connectivity derived from diffusion tensor imaging (DTI) of 22 individuals with left TLE and 39 healthy controls. We validate the model by examining patient-control differences in simulated seizure on set time and network location. We then investigate the potential of the model for surgery prediction by performing in silico surgical resections, removing nodes from patient networks and comparing seizure likelihood post-surgery to pre-surgery simulations. We find that, first, patients tend to transit from nonepileptic to epileptic states more often than controls in the model. Second, regions in the left hemisphere (particularly with in temporal and subcortical regions) that are known to be involved in TLE are the most frequent starting points for seizures in patients in the model. In addition, our analysis also implicates regions in the contralateral and frontal locations which may play a role in seizure spreading or surgery resistance. Finally, the model predicts that patient-specific surgery (resection are as chosen on an individual, model-prompted, basis and not following a predefined procedure) may lead to better outcomes than the currently used routine clinical procedure. Taken together this work provides a first step towards patient specific computational modelling of epilepsy surgery in order to inform treatment strategies in individuals.
Commentary It is well-demonstrated that resective epilepsy surgery for pharmacoresistant mesial temporal lobe epilepsy (mTLE) is superior over medical therapies (1, 2). The term “pharmacoresistant” is defined here as persistent disabling seizures despite at least two appropriate antiepileptic drug trials at maximally tolerated dosages. Decades of peer-reviewed literature validate the probability of favorable postresective seizure outcomes surpassing the efficacy of medical management strategies for pharmacoresistant localizable mTLE. However, long-term postresective outcome data for both lesional and nonlesional TLE reveal that between 30 and 55 percent of patients inevitably experience a recurrence of unprovoked disabling seizures (3). Epilepsy specialists strive to integrate and interpret the many factors for predicting the probability of prolonged seizure freedom following resective epilepsy surgery. However, a large proportion of patients do not fall into straightforward categories regarding postresective outcomes. One proposed strategy is to implement organized scales (or nomograms) where a set of validated classic outcome measures can be used
Epilepsy Currents, Vol. 16, No. 3 (May/June) 2016 pp. 150–152 © American Epilepsy Society
150
to predict with individual specificity, favorable seizure control prior to resection (4–6). The study by Hutchings et al. deviates from the strategy of targeting sublobar regions for TLE resective surgery by modeling axonally connected networks consisting of “critical nodes” responsible for sustaining the unstable epileptogenic network. This computationally intensive algorithm was developed for simulating resection of these critical nodes in a structural connectivity model of a pharmacoresistant epileptogenic brain network. In effect, the term “focal-onset” becomes a misnomer when such a structural “connectome-based” network strategy is used. Patient-specific diffusion tensor imaging (DTI) was used to derive the epileptogenic networks from baseline brain DTI datasets of 22 patients and 39 controls. The algorithm modeled success rates for patients with left TLE epilepsy comparable to the extensive published postresective outcome literature. The model was reported as potentially capable of providing useful patient-specific recommendations for TLE resective surgery. DTI was used to visualize the extent of a white matterconnected epileptogenic circuit with individual specificity. DTI has a long history of detecting seizure-related changes in white matter-associated water diffusion. An evolution of water diffusion-related DTI abnormalities have been wellestablished to occur both long-term as seen interictally, and subacutely during complex partial status epilepticus. Diffusion
Resective Surgery Using Structural Connectivity Modeling
weighted MR imaging (DWI), a predecessor to DTI, measures the magnitude of water movement by applying an additional directional gradient to standard T2-weighted echo planar imaging sequences. These standard imaging sets acquire the data in the three cardinal planes (XYZ) and average the data to get an estimate of the total diffusion within a voxel. The result can be expressed as the apparent diffusion coefficient (ADC). So, the magnitude of diffusion, or mean diffusivity (MD) in DWI is the average of the ADC. DTI is an evolution of this technique, capable of demonstrating seizure-associated changes in the directionality of white matter-associated water diffusion not possible with DWI and fluid attenuation inversion recovery (FLAIR) sequences (7–9). DTI provides the ability to visualize the directionality of water diffusion with respect to axonal bundle orientation. Water diffusion is typically restricted by cell membranes of myelinated neurons, called anisotropic diffusion or fractional anisotropy (FA). This measure represents the directionality of water diffusion. Water does not diffuse across intact axonal fibers but rather along their major axes. So, intact brain white matter promotes anisotropic diffusion, where anisotropy indices tend toward 1. Conversely, isotropic diffusion, where water diffuses in no apparent direction, represents altered or disrupted axonal pathways. In this scenario, the FA index approaches zero. The MD, or trace, is a measurement of the amplitude of the diffusional motion, putatively reflective of cell hydration. Chronic changes seen with DTI following years of focalonset seizures have been well-described in the literature. For example, the usual pattern in mesial temporal sclerosis (MTS) and cortical dysplasia is a reduced FA often with an associated increased MD on the involved side in the hippocampus, amygdala, and insula (8, 10). In support of secondary epileptogenesis in humans, chronic DTI-related changes in TLE—not necessarily associated with MTS—are not confined to the ipsilateral hippocampus alone but are frequently seen contralaterally (11, 12). Hutchings et al. demonstrated the same ipsilateral and contralateral structures as critical nodes in their connectivity model. Intermediate structures connected by bihemispheric axonally communicating pathways have been shown to involve subcortical relay nodes that include the basal ganglia and thalamus. In longstanding TLE, a significantly reduced FA is often seen in the posterior corpus callosum of patients compared with controls. This finding is in contrast to absence epilepsy, where reduced FA is observed in the anterior corpus callosum (13). This latter finding suggests a more distributed epileptogenic network in primary generalized epilepsy compared to more discrete critical neocortical nodes in the networks of focal-onset epilepsy. The findings support the use of corpus callosotomy rather than resective surgery to potentially disrupt the bulk of interhemispheric communication of pharmacoresistant epileptic circuits shared by the frontal lobe and generalized epilepsies. Conversely, subacute transient decreases in FA can be captured postictally following focal-onset seizures. These transient postictal measures can persist for many hours. The author’s modeling algorithm considers the bistable relationship of an epileptogenic network in TLE. That is, the system possesses two co-existing stable states—specifically, a resting
baseline interictal state and a state with seizure-like dynamics. Although Hutchings et al. do not employ DTI in this manner, a so-called interictal DTI can conceivably be acquired for a patient in a static baseline or interictal state. Transient postictal changes in FA and MD can then be acquired for the same patient and compared with interictal baseline DTI measures. A postprocessing technique called “subtracted postictal DTI” (spiDTI) can effectively subtract changes in FA between the two DTI datasets (14). The technique requires that a postictal DTI is acquired within a few hours following a focal-onset seizure, preferably without generalization secondarily. This latter dataset is then subtracted from a resting interictal state DTI dataset following normalization to generate the resulting footprint of the seizure’s propagation pathway. SpiDTI can be used to visualize subacute remnants of the directionality of an epileptogenic circuit recruited by focal-onset seizures. Of note, chronically reduced FA and increased MD have been shown in distant structures such as the cerebellum in both focal-onset and primary generalized epilepsies (13). Moreover, this structure has been consistently associated with both transient alterations in blood flow reported in the ictal single positron emission computed tomography (SPECT) literature as well. The electrophysiological connection of the cerebellum as a deep brain modulator of epileptiform activity is well established. Although such multimodal diagnostic techniques (briefly presented above) implicate the cerebellum in different epilepsy categories, its complete role in the epilepsies remains unknown. Hutchings et al. did not address the cerebellum as a possible critical node in their subject groups. Lastly, although not addressed by the authors, their planning system can be conceivably applied to direct brain modulation therapy. That is, structural connectivity modeling can be used to facilitate optimal placement of U.S. FDA-approved responsive neurostimulation (RNS) system depth electrode leads (NeuroPace Inc, Mountain View, CA) in identified critical nodes of an epileptogenic network. Such a preimplant model can be used to predict modulating the maximal extent of an epileptogenic network, well beyond the depth lead implant sites. Hutchings et al. presents a connectome-based planning model as a presurgical tool used for simulating patient-specific resective surgery for pharmacoresistant TLE. DTI datasets were collected for a homogeneous group of subjects. The authors state that their strategy represents a first step toward designing a new methodology for patient-specific resective surgery planning to improve surgery success rates beyond their current reported numbers. Additionally, such a system can be used not only for resecting critical nodes in TLE but also for strategically targeting such nodes using laser ablation therapy techniques. Obviously, if the model is ever to be employed clinically, next steps must include validating the model in the same patients, and followed long term, to understand postresective seizure outcomes. Regardless, further development of this modeling strategy will potentially provide an innovative platform upon which to augment not only temporal but also extratemporal resective epilepsy surgery planning. by Marvin A. Rossi, MD, PhD
151
Resective Surgery Using Structural Connectivity Modeling
References 1. Engel J Jr, McDermott MP, Wiebe S, Langfitt JT, Stern JM, Dewar S, Sperling MR, Gardiner I, Erba G, Fried I, Jacobs M, Vinters HV, Mintzer S, Kieburtz K; Early Randomized Surgical Epilepsy Trial (ERSET) Study Group. Early surgical therapy for drug-resistant temporal lobe epilepsy: a randomized trial. JAMA 2012;307:922–930. 2. Engel J Jr, Wiebe S, French J, Sperling M, Williamson P, Spencer D, Gumnit R, Zahn C, Westbrook E, Enos B; Quality Standards Subcommittee of the American Academy of Neurology; American Epilepsy Society; American Association of Neurological Surgeons. Practice parameter: temporal lobe and localized neocortical resections for epilepsy: report of the Quality Standards Subcommittee of the American Academy of Neurology, in association with the American Epilepsy Society and the American Association of Neurological Surgeons. Neurology 2003;60:538–547. 3. de Tisi J, Bell GS, Peacock JL, McEvoy AW, Harkness WF, Sander JW, Duncan JS. The long-term outcome of adult epilepsy surgery, patterns of seizure remission, and relapse: a cohort study. Lancet 2011;378:1388–1395. 4. Gracia CG, Yardi R, Kattan MW, Nair D, Gupta A, Nam I, Bingaman W, Gonzalez-Martinez J, Jehi J. Seizure freedom score: a new simple method to predict success of epilepsy surgery. Epilepsia 2014;56:359–365. 5. Jehi L, Yardi R, Chagin K, Tassi L, LoRusso G, Worrell G, Hu W, Cendes F, Morita M, Bartolomei F, Chauvel P, Najm I, Gonzalez-Martinez J, Bingaman W, Kattan MW. Development and validation of nomograms to provide individualised predictions of seizure outcomes after epilepsy surgery: a retrospective analysis. Lancet Neurol 2015;14:283–290.
152
6. Rossi MA. Postresective outcome nomograms: A patient-specific prediction tool for the clinic? Epilepsy Curr 2015;15:1–3. 7. Diehl B, Symms MR, Boulby PA, Salmenpera T, Wheeler-Kingshott CAM, Barker GJ, Duncan JS. Postictal diffusion tensor imaging. Epilepsy Res 2005;65:137–146. 8. Salmenpera TM, Symms MR, Boulby PA, Barker GJ, Duncan JS. Postictal diffusion weighted imaging. Epilepsy Res 2006;70:133–143. 9. Yogarajah M, Duncan JS. Diffusion-based magnetic resonance imaging and tractography in epilepsy. Epilepsia 2008;49:189–200. 10. Assaf BA, Mohamed B, Abou-Khaled KJ, Williams JM, Yazeji MS, Haselgrove J, Faro SH. Diffusion tensor imaging of the hippocampal formation in temporal lobe epilepsy. Brain 2003;24:1857–1862. 11. Arfanakis K, Hermann BP, Rogers BP, Carew JD, Seidenberg M, Meyerand ME. Diffusion tensor MRI in temporal lobe epilepsy. Magn Reson Imaging 2002;20:511–519. 12. Thivard L, Lehericy S, Krainik A, Adam C, Dormont D, Chiras J, Baulac M, Dupont S. Diffusion tensor imaging in medial temporal lobe epilepsy with hippocampal sclerosis. NeuroImage 2005;28:682–690. 13. Yang T, Guo Z, Luo C, Li Q, Yan B, Liu L, Gong Q, Yao D, Zhou D. White matter impairment in the basal ganglia-thalamocortical circuit of drug-naïve childhood absence epilepsy. Epilepsy Res 2012;99:267– 271. 14. Rossi MA, Stebbins G, Murphy C, Greene D, Brinker S, Sarcu D, Tenharmsel A, Stoub T, Stein MA, Hoeppner TJ, Byrne RW, Moseley ME, Bammer RA, Bild S, Dennis J, Arnett N, Balabanov A, Bergen D, Kanner AM, Smith MC. Predicting white matter pathways for direct neurostimulation therapy. Epilepsy Res 2010;91:176–186.
Planning Resective Surgery Using Structural Connectivity Modeling: The Next-Generation Presurgical Evaluation
Predicting Surgery Targets in Temporal Lobe Epilepsy through Structural Connectome Based Simulations. Hutchings F, Han CE, Keller SS, Weber B, Taylor PN, Kaiser M. PLoS Comput Biol 2015;11:e1004642.
Temporal lobe epilepsy (TLE) is a prevalent neurological disorder resulting in disruptive seizures. In the case of drug resistant epilepsy resective surgery is often considered. This is a procedure hampered by unpredictable success rates, with many patients continuing to have seizures even after surgery. In this study we apply a computational model of epilepsy to patient specific structural connectivity derived from diffusion tensor imaging (DTI) of 22 individuals with left TLE and 39 healthy controls. We validate the model by examining patient-control differences in simulated seizure on set time and network location. We then investigate the potential of the model for surgery prediction by performing in silico surgical resections, removing nodes from patient networks and comparing seizure likelihood post-surgery to pre-surgery simulations. We find that, first, patients tend to transit from nonepileptic to epileptic states more often than controls in the model. Second, regions in the left hemisphere (particularly with in temporal and subcortical regions) that are known to be involved in TLE are the most frequent starting points for seizures in patients in the model. In addition, our analysis also implicates regions in the contralateral and frontal locations which may play a role in seizure spreading or surgery resistance. Finally, the model predicts that patient-specific surgery (resection are as chosen on an individual, model-prompted, basis and not following a predefined procedure) may lead to better outcomes than the currently used routine clinical procedure. Taken together this work provides a first step towards patient specific computational modelling of epilepsy surgery in order to inform treatment strategies in individuals.
Commentary It is well-demonstrated that resective epilepsy surgery for pharmacoresistant mesial temporal lobe epilepsy (mTLE) is superior over medical therapies (1, 2). The term “pharmacoresistant” is defined here as persistent disabling seizures despite at least two appropriate antiepileptic drug trials at maximally tolerated dosages. Decades of peer-reviewed literature validate the probability of favorable postresective seizure outcomes surpassing the efficacy of medical management strategies for pharmacoresistant localizable mTLE. However, long-term postresective outcome data for both lesional and nonlesional TLE reveal that between 30 and 55 percent of patients inevitably experience a recurrence of unprovoked disabling seizures (3). Epilepsy specialists strive to integrate and interpret the many factors for predicting the probability of prolonged seizure freedom following resective epilepsy surgery. However, a large proportion of patients do not fall into straightforward categories regarding postresective outcomes. One proposed strategy is to implement organized scales (or nomograms) where a set of validated classic outcome measures can be used
Epilepsy Currents, Vol. 16, No. 3 (May/June) 2016 pp. 150–152 © American Epilepsy Society
150
to predict with individual specificity, favorable seizure control prior to resection (4–6). The study by Hutchings et al. deviates from the strategy of targeting sublobar regions for TLE resective surgery by modeling axonally connected networks consisting of “critical nodes” responsible for sustaining the unstable epileptogenic network. This computationally intensive algorithm was developed for simulating resection of these critical nodes in a structural connectivity model of a pharmacoresistant epileptogenic brain network. In effect, the term “focal-onset” becomes a misnomer when such a structural “connectome-based” network strategy is used. Patient-specific diffusion tensor imaging (DTI) was used to derive the epileptogenic networks from baseline brain DTI datasets of 22 patients and 39 controls. The algorithm modeled success rates for patients with left TLE epilepsy comparable to the extensive published postresective outcome literature. The model was reported as potentially capable of providing useful patient-specific recommendations for TLE resective surgery. DTI was used to visualize the extent of a white matterconnected epileptogenic circuit with individual specificity. DTI has a long history of detecting seizure-related changes in white matter-associated water diffusion. An evolution of water diffusion-related DTI abnormalities have been wellestablished to occur both long-term as seen interictally, and subacutely during complex partial status epilepticus. Diffusion
Resective Surgery Using Structural Connectivity Modeling
weighted MR imaging (DWI), a predecessor to DTI, measures the magnitude of water movement by applying an additional directional gradient to standard T2-weighted echo planar imaging sequences. These standard imaging sets acquire the data in the three cardinal planes (XYZ) and average the data to get an estimate of the total diffusion within a voxel. The result can be expressed as the apparent diffusion coefficient (ADC). So, the magnitude of diffusion, or mean diffusivity (MD) in DWI is the average of the ADC. DTI is an evolution of this technique, capable of demonstrating seizure-associated changes in the directionality of white matter-associated water diffusion not possible with DWI and fluid attenuation inversion recovery (FLAIR) sequences (7–9). DTI provides the ability to visualize the directionality of water diffusion with respect to axonal bundle orientation. Water diffusion is typically restricted by cell membranes of myelinated neurons, called anisotropic diffusion or fractional anisotropy (FA). This measure represents the directionality of water diffusion. Water does not diffuse across intact axonal fibers but rather along their major axes. So, intact brain white matter promotes anisotropic diffusion, where anisotropy indices tend toward 1. Conversely, isotropic diffusion, where water diffuses in no apparent direction, represents altered or disrupted axonal pathways. In this scenario, the FA index approaches zero. The MD, or trace, is a measurement of the amplitude of the diffusional motion, putatively reflective of cell hydration. Chronic changes seen with DTI following years of focalonset seizures have been well-described in the literature. For example, the usual pattern in mesial temporal sclerosis (MTS) and cortical dysplasia is a reduced FA often with an associated increased MD on the involved side in the hippocampus, amygdala, and insula (8, 10). In support of secondary epileptogenesis in humans, chronic DTI-related changes in TLE—not necessarily associated with MTS—are not confined to the ipsilateral hippocampus alone but are frequently seen contralaterally (11, 12). Hutchings et al. demonstrated the same ipsilateral and contralateral structures as critical nodes in their connectivity model. Intermediate structures connected by bihemispheric axonally communicating pathways have been shown to involve subcortical relay nodes that include the basal ganglia and thalamus. In longstanding TLE, a significantly reduced FA is often seen in the posterior corpus callosum of patients compared with controls. This finding is in contrast to absence epilepsy, where reduced FA is observed in the anterior corpus callosum (13). This latter finding suggests a more distributed epileptogenic network in primary generalized epilepsy compared to more discrete critical neocortical nodes in the networks of focal-onset epilepsy. The findings support the use of corpus callosotomy rather than resective surgery to potentially disrupt the bulk of interhemispheric communication of pharmacoresistant epileptic circuits shared by the frontal lobe and generalized epilepsies. Conversely, subacute transient decreases in FA can be captured postictally following focal-onset seizures. These transient postictal measures can persist for many hours. The author’s modeling algorithm considers the bistable relationship of an epileptogenic network in TLE. That is, the system possesses two co-existing stable states—specifically, a resting
baseline interictal state and a state with seizure-like dynamics. Although Hutchings et al. do not employ DTI in this manner, a so-called interictal DTI can conceivably be acquired for a patient in a static baseline or interictal state. Transient postictal changes in FA and MD can then be acquired for the same patient and compared with interictal baseline DTI measures. A postprocessing technique called “subtracted postictal DTI” (spiDTI) can effectively subtract changes in FA between the two DTI datasets (14). The technique requires that a postictal DTI is acquired within a few hours following a focal-onset seizure, preferably without generalization secondarily. This latter dataset is then subtracted from a resting interictal state DTI dataset following normalization to generate the resulting footprint of the seizure’s propagation pathway. SpiDTI can be used to visualize subacute remnants of the directionality of an epileptogenic circuit recruited by focal-onset seizures. Of note, chronically reduced FA and increased MD have been shown in distant structures such as the cerebellum in both focal-onset and primary generalized epilepsies (13). Moreover, this structure has been consistently associated with both transient alterations in blood flow reported in the ictal single positron emission computed tomography (SPECT) literature as well. The electrophysiological connection of the cerebellum as a deep brain modulator of epileptiform activity is well established. Although such multimodal diagnostic techniques (briefly presented above) implicate the cerebellum in different epilepsy categories, its complete role in the epilepsies remains unknown. Hutchings et al. did not address the cerebellum as a possible critical node in their subject groups. Lastly, although not addressed by the authors, their planning system can be conceivably applied to direct brain modulation therapy. That is, structural connectivity modeling can be used to facilitate optimal placement of U.S. FDA-approved responsive neurostimulation (RNS) system depth electrode leads (NeuroPace Inc, Mountain View, CA) in identified critical nodes of an epileptogenic network. Such a preimplant model can be used to predict modulating the maximal extent of an epileptogenic network, well beyond the depth lead implant sites. Hutchings et al. presents a connectome-based planning model as a presurgical tool used for simulating patient-specific resective surgery for pharmacoresistant TLE. DTI datasets were collected for a homogeneous group of subjects. The authors state that their strategy represents a first step toward designing a new methodology for patient-specific resective surgery planning to improve surgery success rates beyond their current reported numbers. Additionally, such a system can be used not only for resecting critical nodes in TLE but also for strategically targeting such nodes using laser ablation therapy techniques. Obviously, if the model is ever to be employed clinically, next steps must include validating the model in the same patients, and followed long term, to understand postresective seizure outcomes. Regardless, further development of this modeling strategy will potentially provide an innovative platform upon which to augment not only temporal but also extratemporal resective epilepsy surgery planning. by Marvin A. Rossi, MD, PhD
151
Resective Surgery Using Structural Connectivity Modeling
References 1. Engel J Jr, McDermott MP, Wiebe S, Langfitt JT, Stern JM, Dewar S, Sperling MR, Gardiner I, Erba G, Fried I, Jacobs M, Vinters HV, Mintzer S, Kieburtz K; Early Randomized Surgical Epilepsy Trial (ERSET) Study Group. Early surgical therapy for drug-resistant temporal lobe epilepsy: a randomized trial. JAMA 2012;307:922–930. 2. Engel J Jr, Wiebe S, French J, Sperling M, Williamson P, Spencer D, Gumnit R, Zahn C, Westbrook E, Enos B; Quality Standards Subcommittee of the American Academy of Neurology; American Epilepsy Society; American Association of Neurological Surgeons. Practice parameter: temporal lobe and localized neocortical resections for epilepsy: report of the Quality Standards Subcommittee of the American Academy of Neurology, in association with the American Epilepsy Society and the American Association of Neurological Surgeons. Neurology 2003;60:538–547. 3. de Tisi J, Bell GS, Peacock JL, McEvoy AW, Harkness WF, Sander JW, Duncan JS. The long-term outcome of adult epilepsy surgery, patterns of seizure remission, and relapse: a cohort study. Lancet 2011;378:1388–1395. 4. Gracia CG, Yardi R, Kattan MW, Nair D, Gupta A, Nam I, Bingaman W, Gonzalez-Martinez J, Jehi J. Seizure freedom score: a new simple method to predict success of epilepsy surgery. Epilepsia 2014;56:359–365. 5. Jehi L, Yardi R, Chagin K, Tassi L, LoRusso G, Worrell G, Hu W, Cendes F, Morita M, Bartolomei F, Chauvel P, Najm I, Gonzalez-Martinez J, Bingaman W, Kattan MW. Development and validation of nomograms to provide individualised predictions of seizure outcomes after epilepsy surgery: a retrospective analysis. Lancet Neurol 2015;14:283–290.
152
6. Rossi MA. Postresective outcome nomograms: A patient-specific prediction tool for the clinic? Epilepsy Curr 2015;15:1–3. 7. Diehl B, Symms MR, Boulby PA, Salmenpera T, Wheeler-Kingshott CAM, Barker GJ, Duncan JS. Postictal diffusion tensor imaging. Epilepsy Res 2005;65:137–146. 8. Salmenpera TM, Symms MR, Boulby PA, Barker GJ, Duncan JS. Postictal diffusion weighted imaging. Epilepsy Res 2006;70:133–143. 9. Yogarajah M, Duncan JS. Diffusion-based magnetic resonance imaging and tractography in epilepsy. Epilepsia 2008;49:189–200. 10. Assaf BA, Mohamed B, Abou-Khaled KJ, Williams JM, Yazeji MS, Haselgrove J, Faro SH. Diffusion tensor imaging of the hippocampal formation in temporal lobe epilepsy. Brain 2003;24:1857–1862. 11. Arfanakis K, Hermann BP, Rogers BP, Carew JD, Seidenberg M, Meyerand ME. Diffusion tensor MRI in temporal lobe epilepsy. Magn Reson Imaging 2002;20:511–519. 12. Thivard L, Lehericy S, Krainik A, Adam C, Dormont D, Chiras J, Baulac M, Dupont S. Diffusion tensor imaging in medial temporal lobe epilepsy with hippocampal sclerosis. NeuroImage 2005;28:682–690. 13. Yang T, Guo Z, Luo C, Li Q, Yan B, Liu L, Gong Q, Yao D, Zhou D. White matter impairment in the basal ganglia-thalamocortical circuit of drug-naïve childhood absence epilepsy. Epilepsy Res 2012;99:267– 271. 14. Rossi MA, Stebbins G, Murphy C, Greene D, Brinker S, Sarcu D, Tenharmsel A, Stoub T, Stein MA, Hoeppner TJ, Byrne RW, Moseley ME, Bammer RA, Bild S, Dennis J, Arnett N, Balabanov A, Bergen D, Kanner AM, Smith MC. Predicting white matter pathways for direct neurostimulation therapy. Epilepsy Res 2010;91:176–186.
