Brain Network Anaylsis Of Temporal Lobe Epilepsy Based On Magnetoencephalography

As a paroxysmal disease,epilepsy usually arises from focal lesions in the brain,which can trigger convulsions over the entire body.Previous brain network studies suggest that the pathological mechanism of epilepsy cannot be fully explained by the existence of focal lesions in the brain and that epilepsy should be considered as a disease of brain network abnormalities,involving both cortical and subcortical changes.To date,there is no clinical technique that can locate the brain abnormalities responsible for the onset of epilepsy as well as its conduction pathways before surgery.Magnetoencephalography(MEG)is a new tool commonly used in the diagnosis and localization of epilepsy.Using this technique,we can obtain the magnetic source images(MSI)which integrate the electromagnetic signals of brain activity with the images of the brain's anatomy and therefore have the advantage of both high temporal resolution(like EEG)and high spatial resolution(like MRI).Using MEG,this study aimed to examine the functional brain network abnormalities in temporal lobe epilepsy patients with different lateralizations and subtypes.Our main findings are as follows:1.We constructed a brain network model based on the imaginary part of coherent vector(coherent network).Coherence analysis can objectively extract the characteristics of the network,and the imaginary part of coherence is insensitive to the effect of volume effect.MEG waveform analysis can directly analyze abnormal epileptic waves.The combination of these two techniques can greatly improve the objectivity and accuracy of the diagnosis for epilepsy.The regions associated with enhanced connectivity were extracted from the coherent network and compared with the locations of equivalent current dipole(ECDs)and the results of synthetic aperture analysis(SAM).Using the electrocorticography(ECoG)/intracranial electroencephalogram(iEEG)as the reference standard,the combination of coherence networks,ECDs,and SAM can significantly improve the accuracy of the localization of the epileptic focus.Meanwhile,compared with healthy controls,patients with temporal lobe epilepsy showed higher ? band connectivity.This study showed that there do exist some abnormalities in the functional brain networks of patients with temporal lobe epilepsy and our findings not only can provide help to the diagnosis for epilepsy,but also could help to locate the exact epileptogenic focus.2.By applying support vector machine(SVM)to features obtained from functional brain networks,we tried to classify different subtypes of patients with temporal lobe epilepsy.Clinically,patients with temporal lobe epilepsy are associated with diverse symptoms,which are speculated to occur as a result of abnormalities in brain activities of widely-distributed brain regions.Hence,investigations of the brain network abnormalities may contribute in understanding the neural basis of these symptoms and in developing effective therapies for this disease.In this study,patients with temporal lobe epilepsy were divided into two subtypes: complex partial seizures(CPS)and simple partial seizures(SPS).The clinical outcomes of these two subtypes are different,whereas it is difficult to detect differences between their origins of electroencephalogram.Based on the constructed brain network,SVM was used to classify the two subtypes,showing a high classification accuracy in our sample.After being re-classified by random sampling,similar results are obtained in multiple highly overlapping brain regions,suggesting the reliability and credibility of our classifier.Our results demonstrate that there are indeed brain network differences between these two subtypes.Given that CPS are clinically accompanied by loss of consciousness and easier to progress to refractory epilepsy,brain network differences between these two subtypes may help to unveil the underlying the neural substrate of the disturbance of consciousness in epilepsy patients and to develop early intervention procedures.3.Classification of lateralization of temporal lobe epilepsy brain network based on magnetic source signals.To solve the lateral localization of temporal lobe epilepsy,eliminate the interference of mirror image foci on epileptic focus,and improve the efficacy of epilepsy surgery,SVM was applied to classify the characteristics of brain magnetic signals in patients with unilateral temporal lobe epilepsy.Our results demonstrated that MEG was an excellent technique for detecting lateralization of temporal lobe epilepsy,and that source spatial brain network parameters can effectively distinguish lateralization.In the clinical diagnosis process,the nodal degree is a best choice for classifying temporal lobe epilepsy and healthy subjects,while the betweenness centrality is a better choice for evaluating the lateralization of temporal lobe epilepsy.Our findings indicate that the combination of MEG and SVM can help to reveal the pathological mechanisms underlying brain network abnormalities in temporal lobe epilepsy,and to improve the treatment plan through classification of lateralization.4.When magnetic source imaging is applied to patients with temporal lobeepilepsy with hippocampal sclerosis,the MEG results are highly consistent with the hippocampal sclerosis area.Moreover,the resection of the location of MSI area exhibits better clinical curative effect.Magnetic resonance spectroscopy(MRS)is used to analyze the neuronal changes of the brain.It was found that the neuronal parameters of patients with no hippocampus sclerosis also show changes in the areas with abnormal brain magnetic signals.Our results indicate that MEG is a reliable and operable diagnostic tool for the localization of epileptogenic zone.In summary,brain network modelling and SVM classification analysis were applied to the MEG data of patients with temporal lobe epilepsy.By systematically examining the diagnosis,localization,classification and clinical practice of temporal lobe epilepsy,it is found that brain network analysis based on MEG data has the potential of revealing the pathological mechanism underlying the abnormalities of brain networks in patients with temporal lobe epilepsy.