OPTIMIZED EDRSN: MRI IMAGES-BASED BRAIN DISEASE CLASSIFICATION USING DEEP LEARNING TECHNIQUES

Classifying brain diseases such as Alzheimer's and brain tumors using brain magnetic resonance imaging (MRI) images is challenging but essential in medical image analysis and healthcare. Deep Learning (DL) and Machine Learning (ML) techniques have shown promise in automating this classification process. To overcome the challenges in classifying brain diseases, we propose a novel DL technique. To classify the brain disease, we use the data from two brain MRI image datasets and perform the noise reduction and normalization on input pictures to reduce the complexity of classification. Then, we retrieve the essential attributes such as shape, position, and size using the EfficientNetV2 technique from the segmented images. Classification is performed using the Enhanced Deep Residual Shrinkage Network (EDRSN) technique and also the Enhanced Northern Goshawk Optimization (ENGO) algorithm to improve the categorization accuracy and evaluation - the evaluation outcomes of the proposed model compared with "state-of-the-art" techniques.