Epilepsy is a recurrent disorder of brain function caused by abnormal excessive and hypersynchronous neuronal discharges,seriously affecting patient's life quality.MRI technologies can noninvasively evaluate changes of brain structure and function,having important value for etiological diagnosis,preoperative evaluation and postoperative follow-up of epilepsy,hence providing strong support for precise individualized diagnosis and treatment of epilepsy.The advancements of MRI technologies in epilepsy were reviewed in this article.

Objective To explore the angioarchitectural characteristics of brain arteriovenous malformations(AVM)associated with epilepsy.Methods Data of conventional MRI,whole brain time of flight MR angiography(TOF MRA)and digital subtraction angiography(DSA)of 70 cases of AVM were retrospectively analyzed.The patients were divided into epilepsy group(n=22)and non-epilepsy group(n=48)based on presence or absence of epileptic seizures,and clinical data,conventional imaging findings and angioarchitectural characteristics of AVM were compared between groups.Multivariate logistic regression analysis was employed to identify independent risk factors of seizures associated with AVM.Results Significant differences of lesion's location,located in functional area or not,as well as of the length of draining vein were found between groups(allP＜0.05).Lesion located in functional area(OR[95％CI]=16.409[2.142,125.733],P=0.007)and longer draining vein length(OR[95％CI]=1.350[1.143,1.594],P＜0.001)were both independent risk factors of AVM related seizures.Conclusion Lesion located in functional area and longer draining vein length were both independent risk factors of AVM related seizures.The length of drainage vein was a simple and repeatable quantitative indicator for identifying AVM at high risk of epilepsy.

Objective To explore the application value of MR MAGiC sequence reconstructed based on deep learning reconstruction(DLR)for localizing epileptogenic focus of intractable epilepsy.Methods Fifty intractable epilepsy patients(epilepsy group)and 20 healthy controls(control group)were retrospectively collected.Brain MR MAGiC sequence was scanned,then traditional reconstruction(inverse Fourier transform)and DLR algorithm were performed to obtain MAGiC and MAGiC DLR images,respectively.The corresponding quantitative parameter maps were generated after post-processing,including T1 mapping,T2 mapping and proton density(PD)mapping.The subjective scores of noise,artifacts,structural clarity and overall quality were compared between two kinds of quantitative parameter images.The asymmetry index(AI)of quantitative parameters(T1,T2 and PD values)between the affected side and the contralateral side in epilepsy group,the contralateral side or the suspected epileptogenic focus in epilepsy group and HC group were calculated.Then epileptogenic focus were localized based on performance of MAGiC and MAGiC DLR corresponding quantitative parameter maps combining obtained AI of quantitative parameter values.According to surgical results,follow-up review,electroencephalogram or PET results,the accuracy of MAGiC and MAGiC DLR images for localizing epileptogenic focus was calculated.Results Compared with MAGiC quantitative parametric maps,noise of MAGiC DLR quantitative parametric maps reduced,while subjective scores of structure clarity and overall quality improved(all P＜0.05).The accuracy of MAGiC and MAGiC DLR images for localizing epileptogenic focus was 82.00％(41/50)and 88.00％(44/50),respectively.Conclusion MR MAGiC sequence based on DLR was beneficial for improving accuracy of localizing epileptogenic focus in intractable epilepsy.

Objective To observe hierarchical differences in brain functional networks in unilateral mesial temporal lobe epilepsy(mTLE)patients with different outcomes after surgery.Methods A total of 69 unilateral mTLE patients who underwent resection of epileptogenic lesion on the affected side were retrospectively enrolled.Based on Engel classification 1 year after surgery,the patients were divided into seizure free(SF)group and non-seizure free(NSF)group.Functional connectivity gradient analysis was employed to extract principal gradient explaining the highest variance on preoperative resting-state functional MRI(rs-fMRI),then the whole-brain gradient characteristics and principal gradient values within specific functional networks were compared between groups.Results Principal gradient connected default mode network(DMN)with sensorimotor network(SMN)along a continuous axis.Compared to SF group,NSF group showed a contracted gradient range at both ends(DMN and SMN)of the functional network and weakened hierarchical differentiation,and principal gradient value of DMN was higher,while that of SMN was lower than those in SF group(both P＜0.05).Conclusion Hierarchical differences in brain functional networks in unilateral mTLE patients with different outcomes after surgery mainly present as enhanced DMN and weakened SMN in NSF ones,and the latter two might serve as important neuroimaging markers for evaluating postoperative seizure recurrence.

Objective To observe the value of neurite orientation dispersion and density imaging(NODDI)for diagnosing unilateral temporal lobe epilepsy(TLE)complicated with hippocampal sclerosis(HS)(TLE-HS).Methods Brain diffusion kurtosis imaging(DKI),3D T1WI and 3D fluid attenuated inversion recovery(FLAIR)sequence images were prospectively collected in 55 patients with unilateral TLE-HS(TLE-HS group)and 55 healthy controls(HC group),and NODDI parameter maps were acquired.The hippocampus NODDI parameters values and FLAIR signal intensity were compared among the affected side,the healthy side in TLE-HS group and HC group,as well as between each two,and their value for diagnosing unilateral TLE-HS were analyzed.Results Significant differences of hippocampus intracellular volume fraction(ICVF),isotropic volume fraction(ISOVF),orientation dispersion index(ODI)values and FLAIR signal intensity were found among the affected side,the healthy side in TLE-HS group and HC group(all P＜0.05).There were significant differences of ICVF,ISOVF and ODI values between the affected side and the healthy side in TLE-HS group(all P＜0.05),of ICVF,ISOVF,ODI values and FLAIR signal intensity between the affected side in TLE-HS group and HC group(all P＜0.05),and of ISOVF values and FLAIR signal intensity between the healthy side in TLE-HS group and HC group(both P＜0.05).The area under the curve(AUC)of ICVF,ISOVF,ODI and their combination for differentiating the affected side and the healthy side of TLE-HS was 0.913,0.712,0.912 and 0.964,for differentiating the affected side of TLE-HS and HC was 0.940,0.822,0.871 and 0.971,respectively,and the combination both had the highest AUC(both P＜0.05).The AUC of ISOVF,ODI and their combination for differentiating the healthy side of TLE-HS and HC was 0.666,0.630 and 0.744,respectively,being not significant different(all P＞0.05).The AUC of FLAIR signal intensity for differentiating the affected side and the healthy side of TLE-HS,the affected side of TLE-HS and HC,the healthy side of TLE-HS and HC was 0.627,0.756 and 0.653,respectively.Conclusion Bilateral hippocampus NODDI parameters were helpful for diagnosing unilateral TLE-HS,which might be superior to commonly used FLAIR sequence.

Ischemic stroke(IS)is the most common type of stroke,and the treatment of acute ischemic stroke(AIS)is highly time-dependent.There is considerable individual heterogeneity in the tolerance of brain tissues to ischemia,and some patients may still be benefitted from treatment beyond conventional time window.The application of multimodal imaging in evaluating the core infarct area and ischemic penumbra enabled individualized treatment decisions guided by"tissue window",hence significantly expanded treatment opportunities for patients beyond conventional time window.The application value of multimodal imaging in screening ones able to be benefitted from treatment beyond conventional time window,treatment decision-making and predicting prognosis of AIS patients beyond conventional time window,as well as the challenges and future directions of multimodal imaging in this field were reviewed in this article.

Objective To observe the volume changes of brain regions and their correlation with cognitive function in temporal lobe epilepsy(TLE)patients.Methods Fifty TLE patients(TLE group)and 34 healthy controls(control group)were prospectively enrolled.Brain 3D-T1WI was collected,brain regions were delineated with automated segmentation technique,and hippocampus was segmented additionally with hippocampal brain template(HBT).The volume of brain regions were compared between groups,and the correlation of brain region volume with Montreal cognitive assessment(MoCA)score were analyzed.Results Compared with control group,in TLE group,bilateral cerebrum white matter,right globus pallidus,left posterior cingulate gyrus,left precentral gyrus,right fusiform gyrus,left cerebellum white matter,corpus callosum,left hippocampus(HBT segmentation and conventional segmentation)and right thalamus volumes decreased significantly(all P＜0.05),while right choroid plexus volume increased significantly(P＜0.05).The volume of left hippocampus(HBT segmentation)and right fusiform gyrus were both positively correlated with MoCA score in TLE group(r=0.335,0.376,P=0.009,0.007).Conclusion Widespread atrophy of brain regions could be seen in TLE patients,which were correlated with cognitive function.

Objective To observe the value of machine learning(ML)models based on brain functional network features combining clinical indicators for predicting postoperative outcomes of patients with drug-resistant mesial temporal lobe epilepsy(DR-mTLE).Methods Totally 84 patients with unilateral DR-mTLE who underwent surgery were retrospectively enrolled and classified into seizure free(SF)group(n=55)and non-seizure free(NSF)group(n=29)according to one-year postoperative follow-up.Clinical data were analyzed to screen independent predictors of postoperative outcomes.Based on brain preoperative resting-state functional MRI,brain functional networks were constructed using graph theory analysis,and 587 features were extracted.Five-fold cross validation was used to divide the data into training set and test set,then the optimal brain functional network features related to postoperative outcomes of DR-mTLE patients were selected.Combining with clinically relevant independent predictors,ML models were constructed using classifiers including Gaussian process(GP),logistic regression(LR),support vector machine(SVM)and quadratic discriminant analysis(QDA),respectively,and the prediction efficacy,calibration and clinical value of each ML model were evaluated.Results Both course of disease and lesion location were clinically relevant independent predictors of postoperative outcome of DR-mTLE patients(OR=0.928,5.710,P=0.010,0.016).Four optimal brain function network features were selected,including betweenness centrality of the third zone of cerebellar vermis,degree centrality of right globus pallidus,nodal efficiency of temporal left inferior temporal gyrus and nodal clustering coefficient of left inferior parietal lobule.The average area under the curve(AUC)of GP,LR,SVM and QDA models in test set was 0.868,0.864,0.875 and 0.870,respectively.Calibration curves and decision curve analysis indicated that each ML model had good calibration and high clinical net benefit.Conclusion ML models based on brain functional network features combining with clinical indicators could be used to effectively predict postoperative outcomes in DR-mTLE patients.

Objective To observe the value of pool sign on T2WI for differentiating solitary brain metastasis(SBM)from solitary mass type high grade glioma(HGG).Methods Totally 120 patients with SBM(SBM group)and 184 patients with solitary mass type HGG(HGG group)confirmed by postoperative pathology or clinical follow-up were retrospectively enrolled.The incidence of pool sign were compared between groups,and the efficacy of pool sign for differentiating SBM from solitary mass type HGG was analyzed.Results Pool sign was noticed in 37 cases(37/120,30.83％)in SBM group and 9 cases(9/184,4.89％)in HGG group,the incidence of pool sign in SBM group was higher than that in HGG group(x2=38.063,P＜0.001).The sensitivity,specificity,positive predictive value and negative predictive value of pool sign for differentiating SBM and solitary mass type HGG was 30.83％(37/120),95.11％(175/184),80.43％(37/46)and 67.83％(175/258),respectively.In SBM group,the incidence of pool sign in cases with primary tumor of adenocarcinoma,squamous cell carcinoma and neuroendocrine carcinoma was 40.79％(31/76),16.67％(3/18)and 12.50％(3/24),respectively,with significant difference(x2=8.911,P=0.012),while compared pairwisely showed significant difference between SBM of adenocarcinoma and neuroendocrine carcinoma(x2=6.505,P=0.011).Conclusion The pool sign on T2WI was helpful for differentiating SBM,especially SBM of adenocarcinoma from solitary mass type HGG.

Objective To observe resting-state functional connectivity(rs-FC)changes of mirror neuron system(MNS)brain regions in male nicotine addicts,and to explore their correlations with the degree of addiction.Methods Twenty-five male nicotine addicts(addiction group)and 27 healthy males(control group)were prospectively recruited,and resting-statefunctional MR scanning was performed.The general information and rs-FC values were compared between groups,and correlations of rs-FC values with Fagerstr?m test for nicotine dependence(FTND)and heaviness of smoking index(HSI)scale scores in addiction group were analyzed.Results No significant difference of age nor education level was found between groups(both P＞0.05).Compared to control group,rs-FC values between left inferior frontal gyrus(IFG.L)and right middle temporal gyrus(MTG.R)as well as right middle occipital gyrus(MOG.R)increased(GRF corrected,voxel-level P＜0.005,cluster-level P＜0.05),while between right fusiform gyrus(FFG.R)and right dorsal cingulate gyrus(DCG.R),right supplementary motor area(SMA.R),left medial superior frontal gyrus(SFGmed.L)and left supplementary motor area(SMA.L)decreased(GRF corrected,voxel-level P＜0.005,cluster-level P＜0.05).In addiction group,rs-FC value between IFG.L and MTG.R was negatively correlated with HSI scale score(r=-0.457,P=0.022),while that between IFG.L and MOG.R was negatively correlated with HSI scale and FTND scores(r=-0.440,-0.458,P=0.028,0.021).Conclusion There were specific changes of rs-FC patterns of MNS brain regions in male nicotine addicts,which were correlated with the degree of addiction.