Search

Purpose Optic pathway gliomas (OPG) are low-grade pilocytic astrocytomas accounting for 3-5% of pediatric intracranial tumors. Accurate and quantitative follow-up of OPG using magnetic resonance imaging (MRI) is crucial for therapeutic decision making, yet is challenging due to the complex shape and heterogeneous tissue pattern which characterizes these tumors. The aim of this study was to implement automatic methods for segmentation and classification of OPG and its components, based on MRI. Methods A total of 202 MRI scans from 29 patients with chiasmatic OPG scanned longitudinally were retrospectively collected and included in this study. Data included T2 and post-contrast T1 weighted images. The entire tumor volume and its components were manually annotated by a senior neuro-radiologist, and inter- and intra-rater variability of the entire tumor volume was assessed in a subset of scans. Automatic tumor segmentation was performed using deep-learning method with U-Net+ResNet architecture. A fivefold cross-validation scheme was used to evaluate the automatic results relative to manual segmentation. Voxel-based classification of the tumor into enhanced, non-enhanced, and cystic components was performed using fuzzy c-means clustering. Results The results of the automatic tumor segmentation were: mean dice score = 0.736 ± 0.025, precision = 0.918 ± 0.014, and recall = 0.635 ± 0.039 for the validation data, and dice score = 0.761 ± 0.011, precision = 0.794 ± 0.028, and recall = 0.742 ± 0.012 for the test data. The accuracy of the voxel-based classification of tumor components was 0.94, with precision = 0.89, 0.97, and 0.85, and recall = 1.00, 0.79, and 0.94 for the non-enhanced, enhanced, and cystic components, respectively. Conclusion This study presents methods for automatic segmentation of chiasmatic OPG tumors and classification into the different components of the tumor, based on conventional MRI. Automatic quantitative longitudinal assessment of these tumors may improve radiological monitoring, facilitate early detection of disease progression and optimize therapy management.

OBJECTIVEExperiential phenomena (EP), such as illusions and complex hallucinations, are vivid experiences created in one’s mind. They can occur spontaneously as epileptic auras or can be elicited by electrical brain stimulation (EBS) in patients undergoing presurgical evaluation for drug-resistant epilepsy. Previous work suggests that EP arise from activation of different nodes within interconnected neural networks mainly in the temporal lobes. Yet, the anatomical extent of these neural networks has not been described and the question of lateralization of EP has not been fully addressed. To this end, an extended number of brain regions in which electrical stimulation elicited EP were studied to test whether there is a lateralization propensity to EP phenomena.METHODSA total of 19 drug-resistant focal epilepsy patients who underwent EBS as part of invasive presurgical evaluation and who experienced EP during the stimulation were included. Spatial dispersion of visual and auditory illusions and complex hallucinations in each hemisphere was determined by calculation of Euclidean distances between electrodes and their centroid in common space, based on (x, y, z) Cartesian coordinates of electrode locations.RESULTSIn total, 5857 stimulation epochs were analyzed; 917 stimulations elicited responses, out of which 130 elicited EP. Complex visual hallucinations were found to be widely dispersed in the right hemisphere, while they were tightly clustered in the occipital lobe of the left hemisphere. Visual illusions were elicited mostly in the occipital lobes bilaterally. Auditory illusions and hallucinations were evoked symmetrically in the temporal lobes.CONCLUSIONSThese findings suggest that complex visual hallucinations arise from wider spread in the right compared to the left hemisphere, possibly mirroring the asymmetry in the white matter organization of the two hemispheres. These results offer some insights into lateralized differences in functional organization and connectivity that may be important for functional mapping and planning of surgical resections in patients with epilepsy.

OBJECTIVEPreoperative localization of seizure onset zones (SOZs) is an evolving field in the treatment of refractory epilepsy. Both magnetic source imaging (MSI), and the more recent EEG-correlated functional MRI (EEG-fMRI), have shown applicability in assisting surgical planning. The purpose of this study was to evaluate the capability of each method and their combination in localizing the seizure onset lobe (SL).METHODSThe study included 14 patients who underwent both MSI and EEG-fMRI before undergoing implantation of intracranial EEG (icEEG) as part of the presurgical planning of the resection of an epileptogenic zone (EZ) during the years 2012–2018. The estimated location of the SL by each method was compared with the location determined by icEEG. Identification rates of the SL were compared between the different methods.RESULTSMSI and EEG-fMRI showed similar identification rates of SL locations in relation to icEEG results (88% ± 31% and 73% ± 42%, respectively; p = 0.281). The additive use of the coverage lobes of both methods correctly identified 100% of the SL, significantly higher than EEG-fMRI alone (p = 0.039) and nonsignificantly higher than MSI (p = 0.180). False-identification rates of the additive coverage lobes were significantly higher than MSI (p = 0.026) and EEG-fMRI (p = 0.027). The intersecting lobes of both methods showed the lowest false identification rate (13% ± 6%, p = 0.01).CONCLUSIONSBoth MSI and EEG-fMRI can assist in the presurgical evaluation of patients with refractory epilepsy. The additive use of both tests confers a high identification rate in finding the SL. This combination can help in focusing implantation of icEEG electrodes targeting the SOZ.

Polymicrogyria (PMG) is a common malformation of cortical development associated with a higher susceptibility to epileptic seizures. Seizures secondary to PMG are characterized by difficult-to-localize cerebral sources due to the complex and widespread lesion structure. Tracing the dynamics of interictal epileptiform discharges (IEDs) in patients with epilepsy has been shown to reveal the location of epileptic activity sources, crucial for successful treatment in cases of focal drug-resistant epilepsy. In this case series IED dynamics were evaluated with simultaneous EEG-fMRI recordings in four patients with unilateral peri-sylvian polymicrogyria (PSPMG) by tracking BOLD activations over time: before, during and following IED appearance on scalp EEG. In all cases, focal BOLD activations within the lesion itself preceded the activity associated with the time of IED appearance on EEG, which showed stronger and more widespread activations. We therefore propose that early hemodynamic activity corresponding to IEDs may hold important localizing information potentially leading to the cerebral sources of epileptic activity. IEDs are suggested to develop within a small area in the PSPMG lesion with structural properties obscuring the appearance of their electric field on the scalp and only later engage widespread structures which allow the production of large currents which are recognized as IEDs on EEG.

Objectives Electroencephalography-correlated functional magnetic resonance imaging (EEG-fMRI) allows imaging of brain-wide epileptic networks, and demonstrates that focal interictal epileptic activity is sometimes accompanied by bilateral functional activations. The corpus callosum (CC) facilitates bilateral spread of epileptic activity and at times targeted surgically for drug-resistant epilepsy (DRE). We hypothesized that focal epileptic networks are more unilateral in patients lacking intact CC. Methods We included focal DRE patients who underwent pre-surgical EEG-fMRI and had CC agenesis (group A, n = 5), patients who previously underwent anterior callosotomy as treatment for drop attacks and continued having seizures (group B, n = 6), and control group of patients with focal epilepsy and intact CC (group C, n = 9). Blood-oxygenation-level-dependent (BOLD) signal maps were generated for interictal epileptic discharges. To quantify bi-hemispheric distribution of epileptic networks, laterality indices were compared between groups. Anatomical and diffusion-weighted imaging demonstrated white matter pathways. Results 96% of studies demonstrated bilateral activations. Laterality indices were similar in groups A and C, whereas group B demonstrated a more bilateral network than group C (p = 0.028). Diffusion-weighted and anatomical imaging showed aberrant white matter pathways and larger anterior commissure in groups A and B. 68% of studies showed maximal activation cluster concordant with the presumed epileptic focus, 28% showed non-maximal activation at presumed focus. Significance Focal epileptic activity is associated with bilateral functional activations despite lack of intact CC, and is associated with stronger contralateral activation in patients after anterior callosotomy compared to controls. These findings disprove our initial hypothesis, and combined with white matter structural imaging, may indicate that the CC is not a sole route of propagation of epileptic activity, which might spread via anterior commissure. Our study demonstrates the utility of EEG-fMRI in assessing epileptic networks and potentially aiding in tailoring surgical treatments in DRE patients with callosal anomalies, and in callosal surgeries.

Objective: Although simultaneous recording of EEG and MRI has gained increasing popularity in recent years, the extent of its clinical use remains limited by various technical challenges. Motion interference is one of the major challenges in EEG-fMRI. Here we present an approach which reduces its impact with the aid of an MR compatible dual-array EEG (daEEG) in which the EEG itself is used both as a brain signal recorder and a motion sensor. Methods: We implemented two arrays of EEG electrodes organized into two sets of nearly orthogonally intersecting wire bundles. The EEG was recorded using referential amplifiers inside a 3 T MR-scanner. Virtual bipolar measurements were taken both along bundles (creating a small wire loop and therefore minimizing artifact) and across bundles (creating a large wire loop and therefore maximizing artifact). Independent component analysis (ICA) was applied. The resulting ICA components were classified into brain signal and noise using three criteria: 1) degree of two-dimensional spatial correlation between ICA coefficients along bundles and across bundles; 2) amplitude along bundles vs. across bundles; 3) correlation with ECG. The components which passed the criteria set were transformed back to the channel space. Motion artifact suppression and the ability to detect interictal epileptic spikes following daEEG and Optimal Basis Set (OBS) procedures were compared in 10 patients with epilepsy. Results: The SNR achieved by daEEG was 11.05 +/- 3.10 and by OBS was 8.25 +/- 1.01 (p

OBJECTIVE Providing a reliable assessment of language lateralization is an important task to be performed prior to neurosurgery in patients with epilepsy. Over the last decade, functional MRI (fMRI) has emerged as a useful noninvasive tool for language lateralization, supplementing or replacing traditional invasive methods. In standard practice, fMRI-based language lateralization is assessed qualitatively by visual inspection of fMRI maps at a specific chosen activation threshold. The purpose of this study was to develop and evaluate a new computational technique for providing the probability of each patient to be left, right, or bilateral dominant in language processing. METHODS In 76 patients with epilepsy, a language lateralization index was calculated using the verb-generation fMRI task over a wide range of activation thresholds (from a permissive threshold, analyzing all brain regions, to a harsh threshold, analyzing only the strongest activations). The data were classified using a probabilistic logistic regression method. RESULTS Concordant results between fMRI and Wada lateralization were observed in 89% of patients. Bilateral and right-dominant groups showed similar fMRI lateralization patterns differentiating them from the left-dominant group but still allowing classification in 82% of patients. CONCLUSIONS These findings present the utility of a semi-supervised probabilistic learning approach for presurgical language-dominance mapping, which may be extended to other cognitive domains such as memory and attention.

EEG-fMRI allows the localization of the hemodynamic correlates of neural activity and has been shown to be useful as a diagnostic tool in pre-surgical evaluation of refractory epilepsy. However, EEG recordings may be highly contaminated by artifacts induced by movements inside the magnetic field thus rendering the scan difficult for interpretation. Existing methods for motion correction require additional equipment or hardware modification. We introduce a simple method for motion artifact detection, the conductive gel bridge sensor (CGBS), easily applicable using the standard setup. We report examples of CGBS use in two patients with epilepsy and demonstrate the method's ability to successfully differentiate between epochs of brain activity and those of movement.

OBJECTIVEResection of intraaxial tumors adjacent to the optic radiation (OR) may be associated with postoperative visual field (VF) deficits. Intraoperative navigation using MRI-based tractography and electrophysiological monitoring of the visual pathways may allow maximal resection while preserving visual function. In this study, the authors evaluated the value of visual pathway mapping in a series of patients undergoing awake craniotomy for tumor resection.METHODSA retrospective analysis of prospectively collected data was conducted in 18 patients who underwent an awake craniotomy for resection of intraaxial tumors involving or adjacent to the OR. Preoperative MRI-based tractography was used for intraoperative navigation, and intraoperative acquisition of 3D ultrasonography images was performed for real-time imaging and correction of brain shift. Goggles with light-emitting diodes were used as a standard visual stimulus. Direct cortical visual evoked potential (VEP) recording, subcortical recordings from the OR, and subcortical stimulation of the OR were used intraoperatively to assess visual function and proximity of the lesion to the OR. VFs were assessed pre- and postoperatively.RESULTSBaseline cortical VEP recordings were available for 14 patients (77.7%). No association was found between preoperative VF status and baseline presence of cortical VEPs (p = 0.27). Five of the 14 patients (35.7%) who underwent subcortical stimulation of the OR reported seeing phosphenes in the corresponding contralateral VF. There was a positive correlation (r = 0.899, p = 0.04) between the subcortical threshold stimulation intensity (3–11.5 mA) and the distance from the OR. Subcortical recordings from the OR demonstrated a typical VEP waveform in 10 of the 13 evaluated patients (76.9%). These waveforms were present only when recordings were obtained within 10 mm of the OR (p = 0.04). Seven patients (38.9%) had postoperative VF deterioration, and it was associated with a length of < 8 mm between the tumor and the OR (p = 0.05).CONCLUSIONSIntraoperative electrophysiological monitoring of the visual pathways is feasible but may be of limited value in preserving the functional integrity of the posterior visual pathways. Subcortical stimulation of the OR may identify the location of the OR when done in proximity to the pathways, but such proximity may be associated with increased risk of postoperative worsening of the VF deficit.

Direct cortical stimulation (DCS) is considered the gold-standard for functional cortical mapping during awake surgery for brain tumor resection. DCS is performed by stimulating one local cortical area at a time. We present a feasibility study using an intra-operative technique aimed at improving our ability to map brain functions which rely on activity in distributed cortical regions. Following standard DCS, Multi-Site Stimulation (MSS) was performed in 15 patients by applying simultaneous cortical stimulations at multiple locations. Language functioning was chosen as a case-cognitive domain due to its relatively well-known cortical organization. MSS, performed at sites that did not produce disruption when applied in a single stimulation point, revealed additional language dysfunction in 73% of the patients. Functional regions identified by this technique were presumed to be significant to language circuitry and were spared during surgery. No new neurological deficits were observed in any of the patients following surgery. Though the neuro-electrical effects of MSS need further investigation, this feasibility study may provide a first step towards sophistication of intra-operative cortical mapping.

Head movements during an MEG recording are commonly considered an obstacle. In this computer simulation study, we introduce an approach, the virtual MEG helmet (VMH), which employs the head movements for data quality improvement. With a VMH, a denser MEG helmet is constructed by adding new sensors corresponding to different head positions. Based on the Shannon's theory of communication, we calculated the total information as a figure of merit for comparing the actual 306-sensor Elekta Neuromag helmet to several types of the VMH. As source models, we used simulated randomly distributed source current (RDSC), simulated auditory and somatosensory evoked fields. Using the RDSC model with the simulation of 360 recorded events, the total information (bits/sample) was 989 for the most informative single head position and up to 1272 for the VMH (addition of 28.6%). Using simulated AEFs, the additional contribution of a VMH was 12.6% and using simulated SEF only 1.1%. For the distributed and bilateral sources, a VMH can provide a more informative sampling of the neuromagnetic field during the same recording time than measuring the MEG from one head position. VMH can, in some situations, improve source localization of the neuromagnetic fields related to the normal and pathological brain activity. This should be investigated further employing real MEG recordings.

Much effort has been devoted to developing analysis methods of subdural electroencephalogram and depth electrode recordings of epileptic patients being evaluated for surgical resection. The general approach is to investigate the brain activity at different locations as recorded by the different electrodes in an attempt to localize the epileptogenic focus or foci. Currently, most of the methods are based on the notion that epileptogenic brain activity is associated with changes in synchronization and in complexity. Here we present a method that is based on the temporal dynamics combined with the spectral distribution of energy in terms of frequency-entropy (FE) templates. The FE templates are based upon maximum information partitioning into a set of frequency bands. The FE template is calculated by wavelet packet decomposition followed by an application of the best basis algorithm minimizing the entropy cost function. A comparison between two FE templates is performed by a special quantitative similarity measure according to the overlap in the partitioning into frequency bands and weighted by the bands' entropy. For localization of the epileptogenic foci, the templates of each electrode during the interictal period are compared with a representative template evaluated from the ensemble of all electrodes during the ictal period. We suggest associating the locations that reveal high template similarity to the ictal template with the epileptogenic foci. To test the method and the underlying assumptions, we perform retrospective analysis of the recorded brain activity, from both grid and depth electrodes, from 11 patients suffering from medically intractable epilepsy. Application of the ictal-interictal FE template similarity analysis revealed regions in the epileptic brain in which the interictal characteristics are highly similar to those of the ictal period. To asses the foci we compared the interictal templates of the different electrodes to each other, forming interelectrode similarity matrices. Investigation of these similarity matrices revealed the existence of a single distinct subcluster of electrodes with high interelectrode similarity in the brain activity of seven patients (type-I activity), and the existence of multiple high interelectrode similarity subclusters in the activity of four patients (type-II activity). Comparisons of the analysis results to the medical presurgical evaluations and the outcomes of the surgical resections suggest that the method may be helpful in the chronic evaluation of the epileptogenic zone before operation, and in some cases (type-I activity) without the need to wait for seizures to occur.