AI-Driven Brain Tumor Classification Using Machine Learning and Feature Extraction Techniques for Real-Time Diagnosis

Speedy correct brain tumor identification stands fundamental for initiating prompt medical diagnosis along with creating efficient treatment strategies. The diagnostic methods of today face time constraints along with errors caused by humans so AI-based tumor detection practices become essential for performing real-time tumor screening. This research uses Magnetic Resonance Imaging (MRI) brain scan images as input data to ensure clinically reliable diagnosis outcomes. The research evaluates the use of Color Layout Filter (CLF) and Edge Histogram Filter (EHF) for feature extraction which blends with Bayes Net (BN) and Gaussian Process and AdaBoost classifiers for tumor recognition assessment. The evaluation measures used accuracy, precision, recall, ROC (Receiver Operating Characteristic) and PRC (Precision-Recall Curve) alongside execution time. The results showed that Gaussian Process proved most effective through its execution of CLF by reaching accuracy level (91.98%), precision (0.93), and recall (0.92) while maintaining 0.01 seconds as execution time. The Bayes Net delivered a notable outcome by producing CLF+Bayes Net which attained 91.26% accuracy and 0.98 ROC. Models that used AdaBoost provided less accurate classification results although they operated with faster computational speeds. The research suggests that Gaussian Process combined with CLF stands as the best model option for real-time brain tumor detection using AI. The study emphasizes the need to pick classification algorithms which combine accuracy levels with computational efficiency for purposes of integrating AI-based diagnostic tools into routine clinical practice.