AUTHOR=Wang Jianping , Li Yongxin , Wang Ya , Huang Wenhua 

TITLE=Multimodal Data and Machine Learning for Detecting Specific Biomarkers in Pediatric Epilepsy Patients With Generalized Tonic-Clonic Seizures

JOURNAL=Frontiers in Neurology

VOLUME=Volume 9 - 2018

YEAR=2018

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2018.01038

DOI=10.3389/fneur.2018.01038

ISSN=1664-2295

ABSTRACT=Previous neuroimaging studies of epilepsy with generalized tonic-clonic seizures (GTCS) mainly focus on adults. However, the neural mechanisms that underling this type of epilepsy remain unclear, especially for children. The aim of the present study was to detect the epilepsy effect on brain of children with GTCS and to investigate whether the changes of brain can be used to discriminate between epilepsy children and healthy children at the level of the individual. To achieve this purpose, we measured gray matter (GM) volume and fractional amplitude of low-frequency fluctuation (fALFF) differences on multimodel magnetic resonance imaging in 14 children with GTCS and 30 age- and gender-matched healthy controls. The patients showed GM volume reduction and fALFF increase in the thalamus, hippocampus, temporal and other deep nuclei. Significant decrease of fALFF was mainly found in default mode network. In addition, epileptic duration was significantly negative related to the GM volumes and significantly positive related to the fALFF value of right thalamus. Support vector machine (SVM) applied to GM volume of right thalamus correctly identified epilepsy children with a statistically significant accuracy of 74.42% (P﹤0.002). SVM applied to fALFF of right thalamus correctly identified epilepsy children with a statistically significant accuracy of 83.72% (P﹤0.002). The consistent neuroimaging results indicated that right thalamus plays an important role to reflect the chronic damaging effect of GTCS epilepsy in children. The longer epilepsy history the children had, the greater GM volume reduction and the greater fALFF increase were detected in their right thalamus. GM volumes and fALFF values in the right thalamus can be identified the children with GTCS from the healthy controls with high accuracy and at an individual subject level. These results are likely to be valuable in explaining the clinical problems and understanding the brain abnormalities underlying this disorder.