A Hybrid Feature Optimization and Deep Learning Model for Brain Tumor Detection

One of the most severe and probably fatal medical disorders, brain tumors must be precisely and quickly detected in order to be adequately treated. An automated method for classifying MRI images and identifying benign or malignant brain tumors presented by this study. This is suggested methodology employs a methodical approach that incorporates several machine learning and image processing approaches. To improve picture quality and eliminate noise, the input MRI images undergo pre-processing using a median filter. The tumor region is subsequently using the assigned clustering technique of Fuzzy C-Means (FCM). Precisely identifying the regions that have been affected. Important texture-based information is extracted by feature extraction using the GLCM, or gray-level co-occurrence matrix. The technique Optimization of Particle Swarms (PSO) is employed to maximize feature selection, guaranteeing that the most pertinent features are used for classification. Lastly, Tumor classification is done using a Deep Convolutional Neural Network (DCNN) classifier, which correctly distinguishes between benign and malignant tumors. The system's Key performance indicators are used to evaluate performance indicators like accuracy, specificity, sensitivity, and precision. The implementation, which is done in MATLAB 2013a, shows how well the suggested approach works to improve diagnostic precision and support medical experts in the investigation of brain tumors. The method's resilience is confirmed by the experimental findings, which show better tumor detection capabilities and good classification performance.