Pixel-based Bayesian classification for meningioma brain tumor detection using post contrast T1-weighted magnetic resonance image

This paper introduces Bayesian approach for automated delineation of meningioma brain tumor using post contrast T1 weighted magnetic resonance image. The proposed framework follows the basis of pixel based classification, combination of two stages; feature extraction followed by learning and classification of pixels into desired classes. Both intensity and texture features are extracted. Thereafter, the pixels corresponding to tumor and non tumor region are classified using feature based Bayesian learning. The performance of the proposed methodology is evaluated. The experimental results show its superiority over linear discriminant analysis (LDA), decision tree (DT), and support vector machine (SVM) classifiers.