An adroit naïve Bayesian based sequence mining approach for prediction of MRI brain tumor image

Magnetic Resonance Imaging (MRI) plays a vital role in clinical imaging, especially in the field of brain tumor prediction. However, the images obtained through MRI scan always contains noises thereby leads to fault predictions. Therefore this article focuses on accurate prediction of brain tumor from MRI images. To remove the noise present in the MRI image, a filter is selected through analyzing five different filters. Noise removal is carried out as preprocessing step. The images obtained from preprocessed step are processed to determine important spatial features through Graytone Spatial Dependence Matrix (GSDM) and Tamura methods. These features are computed to determine a subset of features by joint entropy and GA technique. Resulting features are then manipulated using Naïve Bayesian technique to predict the images as normal or benign or malignant. Experimental results presented in this paper shows that the proposed Naïve Bayesian method performs better in predicting the brain tumor present in a patient´s MRI image.