A Convoluted Hybrid Intelligence Optimized Network Model for Dementia with Lewy Bodies prediction using Brain MRIs

The most common progressive illness among the elderly is dementia, whose course can be stopped or slowed down with early detection.Many machine learning approaches have also been used in research to anticipate dementia and its underlying causes.Numerous learning models based on brain imaging have been developed in the literature to predict dementia.They do, however, confront unique difficulties, such as protracted forecasting periods, inefficiency, complex mathematical models, and a high percentage of inaccurate projections.In this work, we make use of clinical data and Magnetic Resonance Imaging (MRI) with predictive algorithms to classify DLB.This study creates and uses a novel automated methodology called the Convoluted Hybrid Intelligence Optimized Network (CHIO-Net) model to predict and categorize dementia.By choosing the best features from the input images, a unique method called Artificial Levy based Rabbit Optimization (AL-RabO) is implemented in this study to minimize the dimensionality of data.Convolutional Neural Network (CNN), a cutting-edge and modern deep learning model, is used in this work to accurately identify and categorize DLB from MRI brain scans.The Hybrid Hummingbird integrated Boosted Optimization (H2BO) model is utilized to simplify the process and guarantee disease detection performance by optimizing the hyper parameter during classification.The proposed CHIO-Net model achieved an accuracy of 99.5% on benchmark datasets, outperforming conventional models such as ResNet (98.72%) and 3D-CNN (87.62%).The performance outputs and outcomes of the proposed CHIO-Net model are validated and assessed using a range of performance measures and publicly accessible MRI brain datasets.