<![CDATA[Medical personnel struggle to detect brain tumors including glioma and meningioma and pituitary tumors because these tumors display matching MRI features which produces different interpretation results between different observers. The research aims to identify three brain tumor types by analyzing MRI image textures and it assesses different classification methods. The Weka software analyzed 3,900 MRI images through Histogram and Gray Level Co-occurrence Matrix (GLCM) and GLRLMGray Level Run Length Matrix (GLRLM) feature extraction methods before Support Vector Machine (SVM)[SF1.1][A1.2] and Naive Bayes and Multilayer Perceptron and Multiclass Classifier and Random Forest algorithms conducted the classification tasks. The analysis revealed that each tumor pair possesses distinct dominant texture characteristics which consist of standard deviation for glioma–meningioma and energy for glioma–pituitary and homogeneity for meningioma–pituitary. The Random Forest algorithm achieved the best classification results in all experiments because it reached 88.62% accuracy for glioma–meningioma and 96.96% accuracy for glioma–pituitary and 95.92% accuracy for meningioma–pituitary while maintaining high sensitivity and specificity values. The research demonstrates that brain tumor MRI image identification becomes more accurate through the combination of texture features with Random Forest classification methods which leads to better medical diagnostic results.]]>
