An efficient method for MRI brain tumor tissue segmentation and classification using an optimized support vector machine

Brain tumors are abnormal cell growths inside the skull that damage brain cells needed for brain function. The complex structure of the human brain makes it challenging to identify and categorize brain tumors. Nevertheless, segmenting brain tumors in MRI images is difficult due to the wide variation in tumor size, location, and intensity. Specifically, we propose a method with four modules: (i) preprocessing; (ii) decomposition; (iii) segmentation; and (iv) classification—that together overcome these difficulties. Here, we first remove the noise from the MRI brain image using a partial differential equation. Those pre-processed images are fed into the contourlet transform which works on the principle of multiscale decomposition of images. The contourlet transform employs a double filter bank structure comprised of the Laplacian pyramid and a directional filter bank to obtain a sparse representation of the smooth contour of an image. These extracted bands are segmented using a novel Possibilistic Fuzzy C-Means clustering algorithm. Brain tissue portions are finally classified into white matter, grey matter, cerebrospinal fluid, edema, and tumor tissues using an Optimized Support Vector Machine whose parameters are optimally chosen using an Opposition-based Grey Wolf Optimization algorithm. The BraTS2021 and Figshare datasets were used to evaluate the proposed method in terms of sensitivity, specificity, accuracy, PPV, NPV, FPR, and FNR. According to the experimental findings, the proposed methodology is superior to the conventional methods. Overall, the analysis demonstrates that the proposed method is more effective than the alternatives.