In this research work, a new automated system is developed for brain tumor detection by using Magnetic Resonance Imaging (MRI) on the basis of machine learning techniques. The major concerns in the brain tumor detection are time consuming, and the classification accuracy dependsonly on clinician’s experience. To address these issues, a new supervised system is developed for brain tumor detection. In this research study, a new segmentation approach was used for improving the brain tumor detection performance and to diminish the complexity of the system. Initially, Anisotropic Diffusion Filter (ADF) was used as an image pre-processing technique for removing noise from the collected brain image. Then, Berkeley Wavelet Transformation (BWT) was utilized for converting the spatial form of pre-processed MRI image into temporal domain frequency. Besides, Support Vector Machine (SVM) was usedas a classification technique to classify the normal and abnormal regions. SVM classifier effectively diminishes the size of resulting dual issue by developing a relaxed classification error bound. In addition, the undertaken classification approach quickly speed up the training process by maintaining a competitive classification accuracy. From the experimental analysis, the proposed system improved dice coefficient >0.9 compared to the existing systems. The experimental investigation validated and evaluated that the proposed system showed good performance related to the existing systems in light of dice coefficient and accuracy.
Mining Data of Noisy Signal Patterns in Recognition of Gasoline Bio-Based Additives using Electronic Nose