Hybrid Texture Features and Support Vector Machine for Brain Tumor Cataloguing in MRI Images: An Inclusive Investigation

Worldwide, cancer kills more people than any other disease, and it will overtake all others as the top killer in the coming decades. The most effective ways to reduce death from brain cancer have been shown to be early identification and treatment. Recent advances in soft computing and image processing have substantially improved the understanding of medical pictures, which in turn has helped in early diagnosis. The absence of effective treatments, very high tumor heterogeneity, and low survival rates are hallmarks of GBM. Plus, there aren't a ton of MR pictures. It takes a lot of time and has a number of drawbacks to manually analyze these photographs. Maintaining a laser-like focus during the categorization process, which in turn increases the false hit rate, is also very taxing. For this reason, CAD has great promise as an automated approach for MR image analysis. Improving the accuracy of brain tumor categorization using routinely acquired T1 post-contrast MR imaging is the primary goal of this paper, which aims to develop, deploy, and assess a pattern recognition system for this purpose. Automatic brain tumor identification in MRI scans utilizing texture features extracted from the scans using a HTF and SVM is the primary focus of the first piece of research. Various assessment metrics are used to verify the experimental outcomes, including sensitivity, specificity, accuracy, matching index, overlap fraction, and extra fraction. In terms of classification accuracy, HTF with Achieving the greatest accuracy of 98% with GLCM-based vectors of features and 96%, SVM continuously showed the best performance across both conclusion tables When compared with other classifiers, HTF with SVM has a lower error in classification.