Identifying epilepsy based on machine-learning technique with diffusion kurtosis tensor

Juvenile Myoclonic Epilepsy Imaging Endophenotypes and Relationship With Cognition and Resting-State EEG

Structural neuroimaging studies of patients with Juvenile Myoclonic Epilepsy (JME) typically present two findings: 1-volume reduction of subcortical gray matter structures, and 2-abnormalities of cortical thickness. The general trend has been to observe increased cortical thickness primarily in medial frontal regions, but heterogeneity across studies is common, including reports of decreased cortical thickness. These differences have not been explained. The cohort of patients investigated here originates from the Juvenile Myoclonic Epilepsy Connectome Project, which included comprehensive neuropsychological testing, 3 T MRI, and high-density 256-channel EEG. 64 JME patients aged 12-25 and 41 age and sex-matched healthy controls were included. Data-driven approaches were used to compare cortical thickness and subcortical volumes between the JME and control participants. After differences were identified, supervised machine learning was used to confirm their classification power. K-means clustering was used to generate imaging endophenotypes, which were then correlated with cognition, EEG frequency band lagged coherence from resting state high-density EEG, and white and grey matter based spatial statistics from diffusion imaging. The volumes of subcortical gray matter structures, particularly the thalamus and the motor-associated thalamic nuclei (ventral anterior), were found to be smaller in JME. In addition, the right hemisphere (primarily) sulcal pre-motor cortex was abnormally thicker in an age-dependent manner in JME with an asymmetry in the pre-motor cortical findings. These results suggested that for some patients JME may be an asymmetric disease, at least at the cortical level. Cluster analysis revealed three discrete imaging endophenotypes (left, right, symmetric). Clinically, the groups were not substantially different except for cognition, where left hemisphere disease was linked with a lower performance on a general cognitive factor ("g"). HD-EEG demonstrated statistically significant differences between imaging endophenotypes. Tract-based spatial statistics showed significant changes between endophenotypes as well. The left dominant disease group exhibited diffuse white matter changes. JME patients present with heterogeneity in underlying imaging endophenotypes that are defined by the presence and laterality of asymmetric abnormality at the level of the pre-motor sulcal cortex; these endophenotypes are linked to orderly relationships with cognition, EEG, and white matter pathology. The relationship of JME's adolescent onset, age-dependent cortical thickness loss, and seizure upon awakening all suggest that synaptic pruning may be a key element in the pathogenesis of JME. Individualized treatment approaches for neuromodulation are needed to target the most relevant cortical and subcortical structures as well as develop disease-modifying and neuroprotective strategies.

Artificial Intelligence in Pediatric Epilepsy Detection: Balancing Effectiveness With Ethical Considerations for Welfare

BACKGROUND AND AIM

Epilepsy is a major neurological challenge, especially for pediatric populations. It profoundly impacts both developmental progress and quality of life in affected children. With the advent of artificial intelligence (AI), there's a growing interest in leveraging its capabilities to improve the diagnosis and management of pediatric epilepsy. This review aims to assess the effectiveness of AI in pediatric epilepsy detection while considering the ethical implications surrounding its implementation.

METHODOLOGY

A comprehensive systematic review was conducted across multiple databases including PubMed, EMBASE, Google Scholar, Scopus, and Medline. Search terms encompassed "pediatric epilepsy," "artificial intelligence," "machine learning," "ethical considerations," and "data security." Publications from the past decade were scrutinized for methodological rigor, with a focus on studies evaluating AI's efficacy in pediatric epilepsy detection and management.

RESULTS

AI systems have demonstrated strong potential in diagnosing and monitoring pediatric epilepsy, often matching clinical accuracy. For example, AI-driven decision support achieved 93.4% accuracy in diagnosis, closely aligning with expert assessments. Specific methods, like EEG-based AI for detecting interictal discharges, showed high specificity (93.33%-96.67%) and sensitivity (76.67%-93.33%), while neuroimaging approaches using rs-fMRI and DTI reached up to 97.5% accuracy in identifying microstructural abnormalities. Deep learning models, such as CNN-LSTM, have also enhanced seizure detection from video by capturing subtle movement and expression cues. Non-EEG sensor-based methods effectively identified nocturnal seizures, offering promising support for pediatric care. However, ethical considerations around privacy, data security, and model bias remain crucial for responsible AI integration.

CONCLUSION

While AI holds immense potential to enhance pediatric epilepsy management, ethical considerations surrounding transparency, fairness, and data security must be rigorously addressed. Collaborative efforts among stakeholders are imperative to navigate these ethical challenges effectively, ensuring responsible AI integration and optimizing patient outcomes in pediatric epilepsy care.

The value of nomogram based on MRI functional imaging in differentiating cerebral alveolar echinococcosis from brain metastases

OBJECTIVE

This study aims to evaluate the effectiveness of a nomogram model constructed using Diffusion Kurtosis Imaging (DKI) and 3D Arterial Spin Labeling (3D-ASL) functional imaging techniques in distinguishing between cerebral alveolar echinococcosis (CAE) and brain metastases (BM).

METHODS

Prospectively collected were 24 cases (86 lesions) of patients diagnosed with CAE and 16 cases (69 lesions) of patients diagnosed with BM at the affiliated hospital of Qinghai University from 2018 to 2023, confirmed either pathologically or through comprehensive diagnosis. Both patient groups underwent DKI and 3D-ASL scanning. DKI parameters (Kmean, Dmean, FA, ADC) and cerebral blood flow (CBF) were analyzed for the parenchymal area, edema area, and symmetrical normal brain tissue area in both groups. There were 155 lesions in total in the two groups of patients. We used SPSS to randomly select 70% as the training set (108 lesions) and the remaining 30% as the test set (47 lesions) and performed a difference analysis between the two groups. The independent factors distinguishing CAE from BM were identified using univariate and multivariate logistic regression analyses. Based on these factors, a diagnostic model was constructed and expressed as a nomogram.

RESULT

Univariate and multivariate logistic regression analyses identified nDmean1 and nCBF1 in the lesion parenchyma area, as well as nKmean2 and nDmean2 in the edema area, as independent factors for distinguishing CAE from BM. The model's performance, measured by the area under the ROC curve (AUC), had values of 0.942 and 0.989 for the training and test sets, respectively. Calibration curves demonstrated that the predicted probabilities were highly consistent with the actual values, and DCA confirmed the model's high clinical utility.

CONCLUSION

The nomogram model, which incorporates DKI and 3D-ASL functional imaging, effectively distinguishes CAE from BM. It offers an intuitive, accurate, and non-invasive method for differentiation, thus providing valuable guidance for subsequent clinical decisions.

Predictive models of epilepsy outcomes

PURPOSE OF REVIEW

Multiple complex medical decisions are necessary in the course of a chronic disease like epilepsy. Predictive tools to assist physicians and patients in navigating this complexity have emerged as a necessity and are summarized in this review.

RECENT FINDINGS

Nomograms and online risk calculators are user-friendly and offer individualized predictions for outcomes ranging from safety of antiseizure medication withdrawal (accuracy 65-73%) to seizure-freedom, naming, mood, and language outcomes of resective epilepsy surgery (accuracy 72-81%). Improving their predictive performance is limited by the nomograms' inability to ingest complex data inputs. Conversely, machine learning offers the potential of multimodal and expansive model inputs achieving human-expert level accuracy in automated scalp electroencephalogram (EEG) interpretation but lagging in predictive performance or requiring validation for other applications.

SUMMARY

Good to excellent predictive models are now available to guide medical and surgical epilepsy decision-making with nomograms offering individualized predictions and user-friendly tools, and machine learning approaches offering the potential of improved performance. Future research is necessary to bridge the two approaches for optimal translation to clinical care.

[Brain microstructural abnormalities assessed by diffusion kurtosis MRI in patients with focal temporal lobe epilepsy]

OBJECTIVE

To study microstructural abnormalities in epileptogenic focus and in mirror region by diffusion kurtosis (DK) MRI in patients with focal temporal lobe epilepsy.

MATERIAL AND METHODS

The main group included 12 patients (mean age 35 [30.5; 39.0] years, 5 women) with a diagnosis of focal epilepsy in the right temporal lobe (r-TL). The control group consisted of 15 healthy volunteers (mean age 33 [29.0; 39.0] years, 5 women). In both groups all participants underwent clinical and neurologic assessments, EEG and video EEG monitoring, and MRI, including DK MRI.

Evaluation of microstructural changes in the temporal lobes included the study of 10 DK parameters.

RESULTS

In the main group, focal motor/nonmotor or bilateral tonic-clonic seizures were observed in 9 and 11 patients. Ten (83%) patients had focal epileptic activity on routine EEG in the r-TL and 9 (75%) patients had MRI focal changes in the same lobe. There was a significant decrease in fractional (FA) and kurtosis anisotropy (KA), mean (MK), radial (RK), and axial kurtosis (AK), as well as in axonal water fraction (AWF) in the r-TL of patients with epilepsy compared to the control group. Besides, FA, KA, MK, RK, AK, and AWF parameters in the l-TL in the main group were significantly reduced compared to the l-TL in the controls. There was a significant association between the duration of epilepsy, axial and radial extra axonal diffusion as well as axonal diffusivity in the l-TL. No association was observed in the r-TL. In the control group, MK, AK, RK, and AWF positively correlated with age. No correlation between age and MK, AK, RK, and AWF was found in the main group.

CONCLUSION

In patients with an epileptic focus in the r-TL, microstructural changes are observed not only in the area of the epileptic focus, but also in the affected temporal lobe as a whole, and also mirrored in the conditionally healthy temporal lobe. Altered microstructure in the mirror region may reflect changes secondary to bilateral tonic-clonic seizures or evolution to secondary epileptogenic zone. Assessment of DK MRI parameters may provide additional information about the epileptogenic focus and the extent of microstructural abnormalities.

Autism spectrum disorder recognition based on multi-view ensemble learning with multi-site fMRI

Autism spectrum disorders (ASD) is a neurodevelopmental disorder that causes repetitive stereotyped behavior and social difficulties, early diagnosis and intervention are beneficial to improve treatment effect. Although multi-site data expand sample size, they suffer from inter-site heterogeneitys, which degrades the performance of identitying ASD from normal controls (NC). To solve the problem, in this paper a multi-view ensemble learning network based on deep learning is proposed to improve the classification performance with multi-site functional MRI (fMRI). Specifically, the LSTM-Conv model was firstly proposed to obtain dynamic spatiotemporal features of the mean time series of fMRI data; then the low/high-level brain functional connectivity features of the brain functional network were extracted by principal component analysis algorithm and a 3-layer stacked denoising autoencoder; finally, feature selection and ensemble learning were carried out for the above three brain functional features, and a classification accuracy of 72% was obtained on multi-site data of ABIDE dataset. The experimental result illustrates that the proposed method can effectively improve the classification performance of ASD and NC. Compared with single-view learning, multi-view ensemble learning can mine various brain functional features of fMRI data from different perspectives and alleviate the problems caused by data heterogeneity. In addition, this study also employed leave-one-out cross validation to test the single-site data, and the results showed that the proposed method has strong generalization capability, in which the highest classification accuracy of 92.9% was obtained at the CMU site.

A mitochondria cluster at the proximal axon initial segment controls axodendritic TAU trafficking in rodent primary and human iPSC-derived neurons

Loss of neuronal polarity and missorting of the axonal microtubule-associated-protein TAU are hallmarks of Alzheimer’s disease (AD) and related tauopathies. Impairment of mitochondrial function is causative for various mitochondriopathies, but the role of mitochondria in tauopathies and in axonal TAU-sorting is unclear. The axon-initial-segment (AIS) is vital for maintaining neuronal polarity, action potential generation, and—here important—TAU-sorting. Here, we investigate the role of mitochondria in the AIS for maintenance of TAU cellular polarity. Using not only global and local mitochondria impairment via inhibitors of the respiratory chain and a locally activatable protonophore/uncoupler, but also live-cell-imaging and photoconversion methods, we specifically tracked and selectively impaired mitochondria in the AIS in primary mouse and human iPSC-derived forebrain/cortical neurons, and assessed somatic presence of TAU. Global application of mitochondrial toxins efficiently induced tauopathy-like TAU-missorting, indicating involvement of mitochondria in TAU-polarity. Mitochondria show a biased distribution within the AIS, with a proximal cluster and relative absence in the central AIS. The mitochondria of this cluster are largely immobile and only sparsely participate in axonal mitochondria-trafficking. Locally constricted impairment of the AIS-mitochondria-cluster leads to detectable increases of somatic TAU, reminiscent of AD-like TAU-missorting. Mechanistically, mitochondrial impairment sufficient to induce TAU-missorting results in decreases of calcium oscillation but increases in baseline calcium, yet chelating intracellular calcium did not prevent mitochondrial impairment-induced TAU-missorting. Stabilizing microtubules via taxol prevented TAU-missorting, hinting towards a role for impaired microtubule dynamics in mitochondrial-dysfunction-induced TAU-missorting. We provide evidence that the mitochondrial distribution within the proximal axon is biased towards the proximal AIS and that proper function of this newly described mitochondrial cluster may be essential for the maintenance of TAU polarity. Mitochondrial impairment may be an upstream event in and therapeutic target for AD/tauopathy.