Anatomical and radiological evaluation of different regions of the brain in epilepsy patients

Epilepsy is a chronic, heterogeneous, neurological disorder characterized by various motor and cognitive deficits, affecting millions of people of all ages and genders worldwide. The aim of our study is to analyze the anatomical and radiological evaluation of different regions of the brain between children diagnosed with epilepsy and healthy children of the same age without any known health problems by both statistical methods and Machine Learning methods. A total of 179 children aged between 1 and 15 years were included in our study. Among 101 children with epilepsy, 35 (34.65%) were girls and 66 (65.35%) were boys. Among 78 healthy children, 34 (43.59%) were girls and 44 (56.41%) were boys. The mean ages of children with epilepsy and healthy children were 6.91±4.34 years and 7.25±4.25 years, respectively. Epilepsy and healthy control groups were similar regarding male and female distribution (p=0.278) and age (p=0.600). Amygdala, hippocampus, insula, thalamus, caudate nucleus, corpus callosum, and the largest area of the brain parameters were measured higher in healthy children than in children with epilepsy. At the same time, statistically significant differences were found between epilepsy and healthy groups in amygdala, insula, and corpus callosum measurements. Furthermore, in terms of the results obtained, we can say that 77 percent of 100 children will be correctly labeled as having epilepsy if our model is deployed. Automatic prediction of epilepsy cases among children has been studied and based on the results obtained, it is concluded that Machine Learning, when used in combination with feature selection, can produce very satisfactory results in automatic detection of epilepsy cases. Moreover, the anatomical and radiological reference values obtained from this study are important for the evaluation of the dimensions of the structures of both epileptic and healthy children in terms of clinical and pathological processes.