Abstract
Epilepsy disorder characterized by recurrent seizures, which is common in 60%-88% of patients with diffuse lowgrade gliomas, especially those in superficial cortical or insular regions. Understanding the connection between tumor morphology and epileptogenicity helps to refine diagnostic approaches and support therapeutic interventions. Identifying genetic clusters based on individual genetic profiles, supports to improve the epilepsy treatment methods. The study found the volume of white matter, grey matter, and cerebral spinal fluid in relation to epilepsy occurrence and severity. The preprocessing steps of skull stripping, feature scaling by k-means clustering, and radiomic feature selection by logistic regression models were analyzed. The CNN classifier was used to interpret the data to calculate the volumes of white matter (WM), gray matter (GM), and cerebrospinal fluid (CSF) volumes by marching cube algorithm. The performance metrics are calculated by machine learning (ML) classifiers like Support vector machine (95%), Logistic Regression (91%), AdaBoost (89%), Gaussian Naïve Base (87.5%), Gradient Boost (87%), and deep learning (DL) classifiers like CNN (96%) and DNN (79%). The study used classifiers to assess the accuracy and effectiveness of brain structures by prediction models. Although limited by dataset size, it offers valuable insights into epilepsy disorders with radiomic features. Future research should focus on multimodal analysis, and real-time data integration for improved diagnostic. This is the baseline study in the classification of brain tumor epilepsy (BTE) for upcoming research. Over all study aims to quantitatively assess the relationship between brain tumor morphology and epilepsy using deep learning models applied to MRI data.
Keywords: Canny Edge Algorithm, Deep Learning Models, Machine Learning Models, Marching Cube Algorithm, Volumetric Analysis.