KEY WORDS neuronavigation; MRI postprocessing; cryptogenic epilepsy; surgery Epilepsy is one of the most common neurological disorders,

Transcription

1 CLINICAL ARTICLE J Neurosurg 128: , 2018 A multimodal concept for invasive diagnostics and surgery based on neuronavigated voxel-based morphometric MRI postprocessing data in previously nonlesional epilepsy *Daniel Delev, MD, 1 Carlos M. Quesada, MD, 2 Alexander Grote, MD, 1 Jan P. Boström, MD, 1 Christian Elger, MD, 2 Hartmut Vatter, MD, 1 and Rainer Surges, MD 2 Departments of 1 Neurosurgery and 2 Epileptology, University of Bonn, University Medical Center, Bonn, Germany OBJECTIVE Diagnosis and surgical treatment of refractory and apparent nonlesional focal epilepsy is challenging. Morphometric MRI voxel-based and other postprocessing methods can help to localize the epileptogenic zone and thereby support the planning of further invasive electroencephalography (EEG) diagnostics, and maybe resective epilepsy surgery. METHODS The authors developed an algorithm to implement regions of interest (ROI), based on postprocessed MRI data, into a neuronavigation tool. This was followed by stereotactic ROI-guided implantation of depth electrodes and ROI-navigated resective surgery. Data on diagnostic yield, histology, and seizure outcome were collected and evaluated. RESULTS Fourteen consecutive patients with apparently nonlesional epilepsy were included in this study. Reevaluation of the MR images with the help of MRI postprocessing analysis led to the identification of probable subtle lesions in 11 patients. Additional information obtained by SPECT imaging and MRI reevaluation suggested possible lesions in the remaining 3 patients. The ROI-guided invasive implantation of EEG yielded interictal and ictal activity in 13 patients who were consequently referred to resective surgery. Despite the apparently negative MRI findings, focal cortical dysplasia was found in 64% of the patients (n = 9). At the last available outcome, 8 patients (57%) were completely seizure free (International League Against Epilepsy Class 1). CONCLUSIONS The results demonstrate the feasibility and usefulness of a robust and straightforward algorithm for implementation of MRI postprocessing-based targets into the neuronavigation system. This approach allowed the stereotactic implantation of a low number of depth electrodes only, which confirmed the seizure-onset hypothesis in 90% of the cases without causing any complications. Furthermore, the neuronavigated ROI-guided lesionectomy helped to perform resective surgery in this rather challenging subgroup of patients with apparent nonlesional epilepsy. KEY WORDS neuronavigation; MRI postprocessing; cryptogenic epilepsy; surgery Epilepsy is one of the most common neurological disorders, affecting approximately 50 million people worldwide. 30 About 30% of the epilepsies are refractory to medical treatment 13 and are associated with devastating socioeconomic consequences, diminished quality of life, and higher morbidity and mortality rates. 2 Resective surgery is recognized as an effective treatment option for temporal lobe epilepsy (TLE) with proven results better than medical treatment. 5,28 However, in approximately 30% of the patients with refractory focal epilepsy, seizures arise extratemporally (extratemporal lobe epilepsy [ETLE]). 3 The surgical treatment of ETLE often presents a challenge for neurosurgeons and epileptologists that is reflected by moderate seizure-free rates, which are rarely higher than 50%. 4,20 The accurate localization of the epileptogenic zone represents one of the greatest difficulties during the presurgical evaluation of ETLE. 31 Although the routine establishment of high-resolution MRI has revolutionized epilepsy treatment by proper detection of the epileptogenic zone, in approximately 20% 40% of the patients with ETLE no surgical target can be identified. 12 Patients with MRI- ABBREVIATIONS EEG = electroencephalography; ETLE = extratemporal lobe epilepsy; FCD = focal cortical dysplasia; FLAIR = fluid-attenuated inversion recovery; ieeg = invasive EEG; ILAE = International League Against Epilepsy; ISAS = ictal-interictal SPECT analysis by SPM; MAP = morphometric analysis program; ROI = region of interest; rroi = resection ROI; SISCOM = subtraction ictal SPECT coregistered to MRI; SPM = statistical parametric mapping; TLE = temporal lobe epilepsy. SUBMITTED June 27, ACCEPTED December 21, INCLUDE WHEN CITING Published online June 16, 2017; DOI: / JNS * Drs. Delev and Quesada contributed equally to this work AANS 2018, except where prohibited by US copyright law

2 Multimodal concept for surgical treatment of nonlesional epilepsy negative epilepsy have the poorest seizure outcome and represent one of the greatest treatment challenges in tertiary epilepsy centers. 21 Different technical tools, imaging modalities, and improvements have been developed in the hope of facilitating the treatment of MRI-negative epilepsy. 18 One of those image modalities is the computational postprocessing of structural MR images by voxel-based comparison of morphometric changes in the gray/white delineation between patients and healthy individuals (morphometric analysis program [MAP] analysis). 8,9,23 Recent studies have shown that MAP analysis can facilitate and increase the detection rate of focal cortical dysplasia (FCD) in patients with previously diagnosed MRI-negative epilepsy. 23,24 Especially in cases with very small FCDs, the postprocessing results can be transferred to the structural MR image for a second look and can eventually lead to recognition of subtle lesions. 25 However, MAP analyses are commonly performed separately from routine diagnostic tests, often limiting their availability during surgical procedures such as invasive electroencephalography (EEG) or resective surgery. 19 As recently reported, a multimodal concept is commonly required during epilepsy diagnostic tests to create a precise working hypothesis about the localization of the epileptogenic zone. 6,19 Such a multimodal concept should be robust, straightforward, and time efficient to be routinely applicable. Such practicable workflow has already been reported in 1 patient with previously nonlesional epilepsy, in whom electrode implantation was based on the integration of MAP results into the neuronavigation. 26 During the past few years, an optimized and extended workflow was routinely integrated into the presurgical evaluation of patients with apparently nonlesional epilepsy at our center. Even more important, the neuronavigationintegrated data were used to designate a precise resection target, thus facilitating the surgical process. In this study we report on 14 consecutive cases with apparently nonlesional epilepsy who underwent this algorithm and highlight some of the main aspects of the workflow. First, we have improved the integration of MAP (and other functional data) into the neuronavigation by creating regions of interest (ROIs), which were then transferred into the navigation software. Second, those ROIs were confirmed to match the epileptogenic zone by implanting a low number of depth electrodes, thus keeping the complication rate to a minimum. Finally, the ROIs were directly used for resection planning in those patients, who were referred for surgery. The final aim was to validate the use of the algorithm in clinical practice by reporting the detection rate of epileptic EEG activity and histological findings after ROI-navigated surgery for patients with apparently nonlesional epilepsy. Methods Patient Characteristics and Noninvasive Presurgical Evaluation This was a single-center retrospective study. A total of 14 consecutive patients with apparently cryptogenic ETLE, treated from 2012 until 2015, were included and evaluated, which represents about 10% of the patients undergoing presurgical assessment during this time period. All patients underwent comprehensive presurgical workup at the Department of Epileptology and surgical treatment at the Department of Neurosurgery, both at the University Hospital in Bonn. Further patient characteristics are presented in Table 1. This retrospective audit was approved as such by the ethics committee of the University of Bonn Medical Center. For each patient, 3.0-T MRI, ictal and interictal scalp video-eeg, and neuropsychological testing were performed. The epilepsy was defined as cryptogenic if no definitive epileptogenic lesion could be found after conventional visual analysis by an experienced neuroradiologist. Postprocessing was performed using the MAP07 toolboxes for statistical parametric mapping (SPM; Wellcome Department of Cognitive Neurology) and MATLAB (The MathWorks, Inc.). The MAP analysis rationale and methods have been previously described. 8,10 Postprocessing was performed on both 3D isotropic T1- and FLAIRweighted images acquired at 3 T using the commercially available tool from the original developers. The resulting feature z-score maps were visually inspected in search of focal deviations from the norm. Abnormalities that were observed on more than 1 feature map were given more consideration. Back and forth comparison of these results with the structural images and the data collected from the EEG were used for completing the final report as well. In some of the patients, additional ictal subtraction SPECT imaging (interictal SPECT subtracted from ictal SPECT) was performed using the ictal-interictal SPECT analysis by SPM (ISAS) implementation 14 of the usual subtraction ictal SPECT coregistered to MRI (SISCOM) analysis. ROI Creation and Further Image Processing MRIcro ( mricro) and the FSL Suite (FMRIB Analysis Group) were used for evaluation of the results and creation of the ROIs. The ROIs were manually drawn on the T1-weighted images using the MAP feature maps (in most cases the extension map) as a guide. In some cases with highly significant single clusters, a simple threshold of the feature map was enough to obtain the ROI. After the ROIs were superimposed on the original T1-weighted image, the resulting Analyze image was converted to DICOM using the opensource tool kit XMedCon ( net/). Stereotactic Implantation of Depth Electrodes All cases were discussed in regular multidisciplinary conferences. The feasibility of the MAP lesion being the probable epileptogenic zone was considered by crosschecking the semiological and neurophysiological information, other image modalities, as well as a second look at the structural MRI. In some cases the putative structural abnormalities were identified only after automatic postprocessing analysis that guided the neuroradiologists ( second look ). These MRI abnormalities were considered subtle lesions. After a consensus was reached, patients were pro- 1179

3 D. Delev et al. TABLE 1. Patient characteristics Case No. Sex Age (yrs) Surgery Onset MAP ISAS Localization of the Lesion Depth Electrodes Invasive EEG Histology Treatment Follow-Up (mos) Engel Class ILAE (1 6) 1 M NP Frontal 4 Seizure & ID FCD IIA Lesionectomy 12 IA 1 2 M Frontal 2 Seizure & ID FCD IB Lesionectomy 36 IA 1 3 M NP Frontal 2 Seizure & ID FCD IIA Lesionectomy 20 IA 1 4 M Frontal 2 Seizure & ID FCD IIA Lesionectomy 16 IA 1 5 F Frontal 1 Seizure & ID Gliosis Lesionectomy 24 IIIA 4 6 F Frontal 2 Seizure & ID FCD IIB Lesionectomy 18 IA 1 7 M 34 6 Frontal 2 Seizure & ID Gliosis Lesionectomy 16 IA 1 8 M Insular 2 ID Gliosis Lesionectomy 16 IB 2 9 F NP Frontal 2 Seizure Radiosurgery 20 IVB 5 10 F NP Frontal 2 Seizure & ID Unspecific Lesionectomy 16 IIB 3 11 M Frontal 2 Seizure & ID FCD IIB Lesionectomy 16 IIB 3 12 F Frontal 4 Seizure & ID FCD IA Lesionectomy 17 IIA 3 13 M NP Frontal 2 Seizure & ID FCD IIB Lesionectomy 5 IA 1 14 M NP Frontal 2 Seizure & ID FCD IIB Lesionectomy 6 IA 1 + = positive; = negative; ID = interictal discharges; NP = not performed. posed to undergo invasive surgical evaluation to confirm the putative epileptogenic focus by video-eeg recordings. The created ROIs were used as targets for the depth electrodes. Subdural strip or grid electrodes were not used in this series. The T1-weighted DICOM images with the superimposed ROI were easily imported into the neuronavigation system. For the implantation of all depth electrodes, a Leksell stereotactic frame was used, which was placed on patients heads under general anesthesia. A CT scan was then obtained, the images were exported to iplan Net Stereotaxy version 2.3 (BrainLAB), and an image fusion between the CT scan and the MR image with designated ROIs and trajectories was performed to obtain the coordinates of the electrodes. A CT scan was performed immediately after surgery to check the position of the electrodes and rule out bleedings and further complications. The correct localization of the depth electrodes was also confirmed on MRI, usually after a sufficient number of seizures were registered and before the removal of the electrodes. The mean (± SD) duration of the invasive diagnostic was 7.8 ± 2.9 days (range 3 15 days). Invasive EEG Recordings and Video Monitoring One day after electrode implantation, continuous video- EEG recordings were started at the Department of Epileptology. EEG data acquisition was performed with a Micromed System Plus Evolution System (Micromed) using up to 128 channels, a sampling rate of 200 Hz, and a 16-bit analog to digital converter. Data were band-pass filtered between 0.53 and 70. If necessary, antiepileptic drugs were reduced or withdrawn so that occurrence of habitual seizures was facilitated. Resection Planning If the depth electrodes confirmed the hypothesis of the epileptogenic focus, the ROIs were further processed to match an extended lesionectomy (resection ROIs, [rrois]) and transferred again into the neuronavigation. The rroi included the ROI plus a small amount of adjacent healthy cortex to make the resection anatomically coherent (see Figs. 3C and 4C) with the ultimate goal of completely resecting the cortical abnormality. The complete algorithm is shown in Fig. 1. Results Noninvasive Diagnostics, MAP and SISCOM Analysis In 11 (79%) of the 14 apparently MRI-negative patients, a clinically plausible MAP abnormality was found (Fig. 2, Table 1). In 8 of those patients a very subtle MRI lesion was retrospectively identified when refocusing on the MAP abnormality, usually a possible FCD with almost no cortical thickening and very mild transmantle sign (Fig. 3B). SPECT and SISCOM were performed in 8 patients. In 6 (75%) of the 8 patients, a single informative hyperperfusion cluster was found. In 4 patients SISCOM analysis confirmed the MAP findings and additionally helped identify 2 MAP-negative lesions after reevaluation of the hyperperfusion cluster on the structural images (Fig. 2, Table 1). One patient (Case 7, Table 1) ultimately received electrode implantation after a very mild fluid-attenuated inversion recovery (FLAIR) abnormality in the right cingulum was identified. In this particular patient, MAP and SISCOM analysis were negative, but the seizure semiology during video-eeg monitoring with surface EEG was consistent with this hypothesis. This patient was included in the study because both electrode implantation and guided resection were performed according to the algorithm (Fig. 1). Electrode Implantation, Invasive EEG, and Focus Detection Rate All 14 patients underwent implantation of depth elec- 1180

4 Multimodal concept for surgical treatment of nonlesional epilepsy FIG. 1. Detailed description of the algorithm used to create ROIs based on MAP/ISAS data and the implementation of the ROIs into the neuronavigation system, followed by stereotactic (STX) ROI-guided electrode implantation and rroi-navigated resection. The dashed line shows that the creation of the rroi is based on previously created ROIs used for invasive diagnostics. trodes as part of the invasive presurgical diagnostics. A total of 33 ROI-guided depth electrodes were stereotactically implanted. No complications, hemorrhages, or neurological deficits occurred during or after the invasive diagnostics. Postoperative MRI confirmed the correct localization of the depth electrodes in 12 patients. In 1 patient (Case 9), the depth electrodes were not exactly in the ROI but hit its border and showed epileptogenic activity, making it suitable for further diagnostics. The ROI was missed in 1 patient (Case 4, Table 1), which was corrected after FIG. 2. Workflow leading to invasive diagnostics in the study population. 1181

5 D. Delev et al. FIG. 3. Case 11. MRI visualization of important algorithm steps, starting with postprocessing analysis and ending with resection. A: Identification of a lesion based on differences in junction z-score after MAP analysis. B: Corresponding position on the structural MRI, showing a normal finding. C: Creation of the ROI (green) based on postprocessing data followed by ROIguided implantation of the depth electrodes, confirming the epileptogenicity of the lesion. The blue area represents the extended rroi. D: Postoperative MRI showing extent of the resection. Figure is available in color online only. performing a second implantation procedure (2 additional depth electrodes). The invasive EEG (ieeg) registered interictal and ictal epileptiform discharges in 12 patients. One patient (Case 8, Table 1) presented only interictal discharges and another (Case 9, Table 1) displayed only ictal discharges without interictal activity. Surgery, Histology, and Outcome A total of 13 patients underwent resective surgery after the invasive diagnostics, leading to 12 frontal and 1 insular lesionectomies. The extension of the resection volume 1182 (rroi) was discussed and designated in a multidisciplinary conference (Fig. 3C and D; Fig. 4C and D). One patient underwent radiosurgery, as the rroi would have included the precentral gyrus, thus bearing a high risk for postoperative neurological deficit. The histological analysis findings were distributed as follows: 9 patients with FCD (FCD IA = 1, FCD IB = 1, FCD IIA = 3, FCD IIB = 4), 3 patients with gliosis and 1 patient with unspecific findings. Surgical complications after resective surgery occurred in 3 patients; 1 of them suffered from minor permanent neurological deficit. Mean postoperative follow-up was 17 ± 6.9 months

6 Multimodal concept for surgical treatment of nonlesional epilepsy FIG. 4. Case 2. Visualization of important algorithm steps, starting with ISAS analysis and ending with resection. A: Identification of a lesion based on hyperperfusion during the ISAS analysis. B: Second look on the normal MRI leading to identification of a very subtle lesion (putative FCD). C: Creation of the ROI (green area) based on data from the ISAS analysis followed by ROI-guided implantation of the depth electrodes, confirming the epileptogenicity of the lesion. Creation of the rroi (blue area). D: Postoperative MRI showing extent of the resection. Figure is available in color online only. (range 5 36 months). Postoperative seizure outcome was rated according to both the Engel7 and the International League Against Epilepsy (ILAE)29 classifications (Table 1). At the last available follow-up, 57% of the patients (n = 8) were completely seizure free (ILAE Class 1 or Engel Class IA). The remaining 6 patients continued having seizures. Subsequent analysis to explain surgical failure revealed incomplete (or not extended) resection, unspecific histology, and lack of ictal discharges during the ieeg (Table 2). One patient with Engel Class IV/ ILAE Class 5 (Case 9) underwent radiosurgery because of the eloquently localized epileptogenic lesion. Another patient (Case 8) was seizure free for 1 year but presented with relapse of seizures after discontinuing the antiepileptic drugs. Discussion In this retrospective case series, an algorithm for ROIguided stereotactic implantation of depth electrodes followed by neuronavigated resective surgery was rigorously applied to 14 consecutive patients with previously nonlesional epilepsy, achieving full seizure control in 57% of the cases. The results support the feasibility and usefulness of a technical method for implementation of ROI-based targets into the neuronavigation. The very selective ROIguided stereotactic implantation of a low number of depth electrodes was able to confirm the seizure-onset hypothesis and lead to surgery in 13 of 14 patients. In our opinion, one of the greatest challenges of daily practice during the preparation of complex epilepsy cases for surgery is 1183

7 D. Delev et al. TABLE 2. Failure analysis of the patients who did not become seizure free due to surgical and/or epileptological reasons Case No. Surgical Reasons Epileptological Reasons Resection Histology Invasive EEG 5 Complete and extended resection Unspecific* Seizure & ID 8 Incomplete resection* Unspecific* ID* 9 No resection, only radiotherapy* No histology* Seizure & ID 10 Complete and extended resection Unspecific* Seizure & ID 11 Complete, but not extended resection* Specific (FCD IIB) Seizure & ID 12 Complete, but not extended resection* Specific (FCD IA) Seizure & ID * Characteristics considered to negatively influence the final seizure outcome. to find an appropriate and efficient way of communication and logistics between epileptologists and neurosurgeons. In our case series, we illustrated how such difficulties can easily be overcome: the epileptologist creates an ROI that summarizes all the information required to confirm the seizure-onset hypothesis. By simply implementing the ROI into the neuronavigation system, the neurosurgeon is able to plan independently the implantation procedure and the following resection. Because the ROI implementation into the neuronavigation has made the neurosurgical planning more practical and straightforward, this simple algorithm is currently used on a routine daily basis in our department. Finally, the multimodal concept has been further developed to plan an ROI-navigated lesionectomy, facilitating surgery in a rather challenging subgroup of patients with MRI-negative ETLE. Importance of Stereotactic ROI-Guided Invasive Diagnostic Invasive diagnostics with depth electrodes involves hazards but can hardly be skipped in formerly nonlesional cases. The ROI-guided invasive diagnostics allowed selective targeting, leading to implantation of a very low number of electrodes (mostly 1 or 2). There were no complications related to the electrode implantation. The epileptogenicity of the presumed lesion could be confirmed not only in the 11 cases that were MAP-positive, but also in those with just presumed abnormalities based on SPECT findings or suspected subtle MRI lesions (and consistent electroclinical findings on video-eeg recordings with surface electrodes). Thus, the implantation of a minimal number of depth electrodes facilitated the invasive diagnostic without leading to any obvious disadvantages for the patient. Histological Findings, Seizure Outcome, and Failure Analysis Unspecific histological results or gliosis were found in only 4 patients. These findings are similar to the results recently reported by Wang et al., who described approximately 30% pathologically negative results (gliosis, hamartia, or normal). 24 Eight patients (57%) were seizure free at the last available outcome, which is similar to other case series with ETLE. 4 If compared with series describing MRI-negative extratemporal epilepsies, the seizure-free rate among our cohort was higher, 1,21,22 thus supporting the role of the presented algorithm in creating a constrained focus hypothesis and facilitating the resection. However, this could be also due to the fact that the series of MRI-negative patients with ETLE included very different epilepsy syndromes and histological findings and were not focused on a small and homogeneous group of epilepsy syndromes, mostly caused by FCD, as was the case in our series. Failure analysis of the patients who did not become seizure free suggests that incomplete or narrow resection and unspecific histology may predict unfavorable outcome following resective surgery. Additionally, patients without completely coherent ieeg findings (i.e., lack of ictal activity during ieeg) appear to be at greater risk of not becoming seizure free. Comparison With Previous Studies In their recent studies, Rodionov et al. 19 and Nowell et al. 17 reported a workflow describing the feasibility of 3D neuroimaging during the implantation of intracranial electrodes. There are, however, some differences to our approach. First, the surgical procedures of this study were mainly based on the data delivered from MAP analysis in nonlesional epilepsy. Second, these 2 studies used a standalone software platform (Amira) to integrate the multimodal data, whereas we created a MAP/SISCOM-based ROI, which was then transferred into DICOM format. This allowed the import and coregistration of the ROIs more easily to the structural MR images of the patients, followed by the direct planning of the trajectories for the depth electrodes. A further advantage of the direct ROI integration into the neuronavigation was its availability during the planning and performing of the resection. This last step of the algorithm has been described here for the first time. Two other studies 6,11 have investigated the role of multimodal imaging for apparently nonlesional epilepsy. However, these works predominantly addressed presurgical evaluation and focus localization and not the intraoperative usage of the data. The importance of the intraoperative usage of multimodal data during epilepsy surgery has been described in previous studies by Wellmer et al. 27 and Murphy et al., 16 but these case series did not include any postprocessing data. In contrast to previous works, 15 only a small number of depths electrodes were implanted in our patients, and almost all patients underwent resective surgery after the invasive diagnostic. This fact probably reflects the gono-go mentality of our approach. That is, our approach 1184

8 Multimodal concept for surgical treatment of nonlesional epilepsy aimed at confirming a subtle suspected lesion as being the epileptogenic focus, to proceed with further resection. We intentionally avoided a more extensive exploratory approach, requiring numerous intracranial electrodes. Thus, the risk related to the multiple depth electrode implantations was decreased by simultaneously increasing the rate of successful resective surgery. Our algorithm is somewhat more conservative, as we prefer to use only a small number of electrodes based on MRI or other modalities, thus increasing the number of confirmed hypotheses for resection. Nevertheless, the aim of this study was not to assess the efficacy of this proposed algorithm and to compare it to other approaches, but rather to describe the procedure and to report its feasibility. Limitations of the Study The present study was not designed to evaluate the effectiveness of the MAP analysis for detecting FCDs or to compare the efficacy of resection techniques. Due to the low number of implanted electrodes, the true extent of the seizure generator could have been underestimated due to spatial sampling error. This would also be true if extensive mapping of eloquent cortex would be needed. In such cases, either the implantation of more depth electrodes or the combination of depth and subdural grid electrodes can be taken into consideration. The postoperative seizure outcome, however, was excellent in 57% of our patients, strengthening the validity of our approach. The relatively small number of patients and the retrospective nature of our case series do not allow the identification of selection criteria for suitable candidates and predictors of favorable postsurgical outcome. This is also the reason why the reported seizure-free rate, which was higher than other ETLE studies, should be carefully interpreted and needs validation in a larger cohort. Further development of MRI (greater field strengths, novel technical sequences) and improved postprocessing techniques will hopefully increase the surgical options for patients without lesions. A further potential technical limitation of our approach is the usual shift of the brain caused by the implantation of the depth electrodes, which may impair the accuracy of the overlay of the images before and after the implantation. In view of the low number of implanted electrodes, this shift is likely to be less important than in cases of extended implantations. Conclusions We present an algorithm for invasive diagnostics followed by resective surgery in previously nonlesional epilepsy based on the neuronavigated ROIs obtained after postprocessing of MRI data. Our results support the role of the postprocessing analysis in localizing the epileptogenic zone and suggest that its confirmation can be achieved with a restricted number of depth electrodes. Furthermore, the integration of the ROIs into the neuronavigation facilitated the surgical procedure, which can be a challenging obstacle in cases of apparently nonlesional epilepsy. Finally, all steps were incorporated in a robust and straightforward algorithm, allowing its use in daily clinical practice. References 1. Ansari SF, Tubbs RS, Terry CL, Cohen-Gadol AA: Surgery for extratemporal nonlesional epilepsy in adults: an outcome meta-analysis. Acta Neurochir (Wien) 152: , Baker GA, Nashef L, van Hout BA: Current issues in the management of epilepsy: the impact of frequent seizures on cost of illness, quality of life, and mortality. Epilepsia 38 (Suppl 1):S1 S8, Cascino GD: Surgical treatment for epilepsy. Epilepsy Res 60: , Cascino GD: Surgical treatment for extratemporal epilepsy. Curr Treat Options Neurol 6: , Clusmann H, Kral T, Schramm J: Present practice and perspective of evaluation and surgery for temporal lobe epilepsy. Zentralbl Neurochir 67: , Doelken MT, Mennecke A, Huppertz HJ, Rampp S, Lukacs E, Kasper BS, et al: Multimodality approach in cryptogenic epilepsy with focus on morphometric 3T MRI. J Neuroradiol 39:87 96, Engel J Jr, Van Ness PC, Rasmussen TB, Ojemann LM: Outcome with respect to epileptic seizures, in Engel J Jr (ed): Surgical Treatment of the Epilepsies, ed 2. New York: Raven Press, 1993, pp Huppertz HJ, Grimm C, Fauser S, Kassubek J, Mader I, Hochmuth A, et al: Enhanced visualization of blurred graywhite matter junctions in focal cortical dysplasia by voxelbased 3D MRI analysis. Epilepsy Res 67:35 50, Huppertz HJ, Wellmer J, Staack AM, Altenmüller DM, Urbach H, Kröll J: Voxel-based 3D MRI analysis helps to detect subtle forms of subcortical band heterotopia. Epilepsia 49: , Kassubek J, Huppertz HJ, Spreer J, Schulze-Bonhage A: Detection and localization of focal cortical dysplasia by voxelbased 3-D MRI analysis. Epilepsia 43: , Kurian M, Spinelli L, Delavelle J, Willi JP, Velazquez M, Chaves V, et al: Multimodality imaging for focus localization in pediatric pharmacoresistant epilepsy. Epileptic Disord 9:20 31, Kutsy RL: Focal extratemporal epilepsy: clinical features, EEG patterns, and surgical approach. J Neurol Sci 166:1 15, Kwan P, Brodie MJ: Effectiveness of first antiepileptic drug. Epilepsia 42: , McNally KA, Paige AL, Varghese G, Zhang H, Novotny EJ Jr, Spencer SS, et al: Localizing value of ictal-interictal SPECT analyzed by SPM (ISAS). Epilepsia 46: , Morace R, Di Gennaro G, Picardi A, Quarato PP, Sparano A, Mascia A, et al: Surgery after intracranial investigation with subdural electrodes in patients with drug-resistant focal epilepsy: outcome and complications. Neurosurg Rev 35: , Murphy MA, O Brien TJ, Morris K, Cook MJ: Multimodality image-guided surgery for the treatment of medically refractory epilepsy. J Neurosurg 100: , Nowell M, Rodionov R, Zombori G, Sparks R, Winston G, Kinghorn J, et al: Utility of 3D multimodality imaging in the implantation of intracranial electrodes in epilepsy. Epilepsia 56: , Pardoe H, Kuzniecky R: Advanced imaging techniques in the diagnosis of nonlesional epilepsy: MRI, MRS, PET, and SPECT. Epilepsy Curr 14: , Rodionov R, Vollmar C, Nowell M, Miserocchi A, Wehner T, Micallef C, et al: Feasibility of multimodal 3D neuroimaging to guide implantation of intracranial EEG electrodes. Epilepsy Res 107:91 100, Roper SN: Surgical treatment of the extratemporal epilepsies. Epilepsia 50 (Suppl 8):69 74, Téllez-Zenteno JF, Hernández Ronquillo L, Moien-Afshari F, 1185

9 D. Delev et al. Wiebe S: Surgical outcomes in lesional and non-lesional epilepsy: a systematic review and meta-analysis. Epilepsy Res 89: , Vézina LG: MRI-negative epilepsy: protocols to optimize lesion detection. Epilepsia 52 (Suppl 4):25 27, Wagner J, Weber B, Urbach H, Elger CE, Huppertz HJ: Morphometric MRI analysis improves detection of focal cortical dysplasia type II. Brain 134: , Wang ZI, Jones SE, Jaisani Z, Najm IM, Prayson RA, Burgess RC, et al: Voxel-based morphometric magnetic resonance imaging (MRI) post-processing in MRI-negative epilepsies. Ann Neurol 77: , Wang ZI, Jones SE, Ristic AJ, Wong C, Kakisaka Y, Jin K, et al: Voxel-based morphometric MRI post-processing in MRInegative focal cortical dysplasia followed by simultaneously recorded MEG and stereo-eeg. Epilepsy Res 100: , Wellmer J, Parpaley Y, von Lehe M, Huppertz HJ: Integrating magnetic resonance imaging postprocessing results into neuronavigation for electrode implantation and resection of subtle focal cortical dysplasia in previously cryptogenic epilepsy. Neurosurgery 66: , Wellmer J, von Oertzen J, Schaller C, Urbach H, König R, Widman G, et al: Digital photography and 3D MRI-based multimodal imaging for individualized planning of resective neocortical epilepsy surgery. Epilepsia 43: , Wiebe S, Blume WT, Girvin JP, Eliasziw M: A randomized, controlled trial of surgery for temporal-lobe epilepsy. N Engl J Med 345: , Wieser HG, Blume WT, Fish D, Goldensohn E, Hufnagel A, King D, et al: ILAE Commission Report. Proposal for a new classification of outcome with respect to epileptic seizures following epilepsy surgery. Epilepsia 42: , World Health Organization: Epilepsy fact sheet. ( who.int/mediacentre/factsheets/fs999/en/) [Accessed February 23, 2017] 31. Yun CH, Lee SK, Lee SY, Kim KK, Jeong SW, Chung CK: Prognostic factors in neocortical epilepsy surgery: multivariate analysis. Epilepsia 47: , 2006 Disclosures Dr. Elger has received support in the form of honoraria and consultation fees from Bial, Eisai, Novartis, Pfizer, Desitin, and UCB; he also received federal funding from the DFG (Deutsche Forschungsgemeinschaft). Dr. Surges has served as a consultant to Bial and has received speakers fees from Cyberonics, Eisai, Novartis, and UCB Pharma. Author Contributions Conception and design: Delev, Quesada, Surges. Acquisition of data: Delev, Quesada, Grote, Boström. Analysis and interpretation of data: Delev, Quesada, Surges. Drafting the article: Delev, Quesada, Surges. Critically revising the article: Grote, Boström, Elger, Vatter, Surges. Supplemental Information Previous Presentations Part of this work was presented in abstract form at the 67th annual meeting of the German Society of Neurosurgery (DGNC), Frankfurt, Germany, in June Current Affiliations Dr. Delev: Department of Neurosurgery, University of Freiburg, Germany. Correspondence Daniel Delev, Department of Neurosurgery, University of Freiburg, University Medical Centre, Breisacherstr. 64, Freiburg, Germany. delev.daniel@gmail.com. 1186