Predictive and State-of-the-art Machine Learning Models for Brain Tumors Prediction and Classification

Abstract

Brain tumors as part of complicated diseases impacting delicate functions of the brain, are generally caused by brain cell’s uncontrolled proliferation or within locations such as the pituitary gland, the pineal gland, nerves, and the membranes covering the brain surface. Brain tumors are classed as critical diseases due to the brain’s complex structure, and challenges related to early detection. In recent decades, machine learning (ML) has revolutionized medical imaging processing, innovating how healthcare professionals diagnose diseases and treat various conditions. This technique allows computers to learn to identify specific image features and patterns, and automatically classify different medical conditions. Several ML techniques are commonly used for image preprocessing in diverse domains.

This study aims to predict and classify brain tumors and analyze the performance of ML models: Support Vector Machine (SVM), XGBoost classifier, Light GBM (LGBM), and Random Forest (RF), using performance metrics: precision, recall, f1-score, and accuracy. This research paper reveals that the LGBM classifier outperforms other algorithms, achieving 91% accuracy.

Conclusively, this study illustrated that ML methods prove efficiency in brain tumor prediction at an early stage and provide a robust model for medical professionals to save patient lives.