Brain tumor classification utilizing Triple Memristor Hopfield Neural Network optimized with Northern Goshawk Optimization for MRI image

Brain tumor classification plays a significant role in exact detection of abnormal brain tissues and facilitates the clinical diagnosis of patients. The computer vision researchers have created numerous algorithms, but they still suffer from low accuracy. Therefore, a Brain Tumor Classification utilizing Triple Memristor Hopfield Neural Network optimized with Northern Goshawk Optimization is proposed in this paper for MRI Image (BTC-TMHNNNGOA). Here, the input imageries are pre-processed to increase the images quality using Median Modified Weiner Filtering (MMWF) method. The pre-processed image is supplied to Synchro Extracting Chirplet Transform (SECT) to extract the features of anti-noise interference capability and image resolution. Then the Triple Memristor Hopfield Neural Network (TMHNN) for classifying Tumor and Non-Tumor Image. Afterward, Northern Goshawk Optimization Algorithm (NGOA) is used to optimize the weight parameters of TMHNN classifier for precise classification. The proposed BTC-TMHNN-NGOA technique is activated in MATLAB under metrics, like precision, accuracy, sensitivity, specificity, ROC, F1-score, computational time and computation cost. The proposed method attains 13.88%, 8.75%, and 8.46% higher accuracy for tumor; 9.47%, 14.51% and 10.23% higher accuracy for Non-Tumor on MRI brain image dataset compared with existing methods like Automated Brain Tumor Categorization utilizing Optimized Hybrid Neural Network (EABTC-OHNN), Convolutional Neural Network for MRI-Based Brain Tumor Categorization (CNN-MRI-BTC), Enhancing Convolutional Neural Network using Hybridized Elephant Herding Optimization approach for MRI Categorization Glioma Brain Tumor Grade (OCNN-HEHOA-MRIC-GBTG). The comparative results exemplify the effectiveness of the proposed method and underscore its advantages in automating brain tumor classification.