Gaussian kernelized transformer learning model for brain tumor risk factor identification and disease diagnosis

Brain tumor is an abnormal growth of cells in brain or central spinal canal. Tumors are benign (non-cancerous) or malignant (cancerous). They can be invented in the brain (primary tumors) or spread to other elements of the body. Early detection significantly improves survival rates and overall prognosis for patients by enabling intervention before tumors grow larger or spread, which can complicate treatment. Early intervention also preserves brain function and quality of life while minimizing severe neurological damage or symptoms. Epileptic seizures are a significant clinical symptom and a potential early indicator of brain tumors in patients. Conventional machine learning and deep learning face significant challenges in accurately detecting brain tumors with minimal time consumption. In this paper, a novel technique called multivariate relief matching gaussian kernelized transformer learning (MRMGKTL) model has been developed. The major intent of the MRMGKTL model is to enhance the accuracy of brain tumor detection through the recognition of epileptic seizures. MRMGKTL model comprises data acquisition, feature selection, and classification. In the data acquisition phase, electrical activity data from the brains of patients are collected from datasets for diagnosing brain tumors based on epileptic seizures. Following data acquisition, Sokal– Michener’s multivariate relief matching technique is used to choose the most significant aspects of the dataset. The feature selection process in the proposed method aims to minimize the time required for tumor detection. Using the selected features, brain tumor disease diagnosis is performed using a Gaussian Kernelized transformer learning model to detect and diagnose brain tumors associated with epileptic seizure severity levels with higher accuracy. This approach ensures the accurate identification of brain tumors and associated risk factors with minimal time consumption. Experimental assessment evaluates various factors. Analyzed outcomes demonstrate that the proposed MRMGKTL model achieves superior performance in accuracy of brain tumor diagnosis and reduces time consumption compared to conventional deep learning methods.