Novel Activation Function Design for Multiple Sclerosis Detection in Medical Imaging

The central nervous system is impacted by multiple sclerosis (MS), a chronic and frequently disabling illness that causes a variety of physical and mental disabilities. Effective management and therapy depend heavily on early diagnosis and the ability to anticipate the course of the disease. The use of artificial intelligence (AI) approaches to forecast the onset and course of MS is investigated in this work. The paper introduces a novel function called Gelu Error Activation Function (GERF), that computes Gauss error function (erf) of Gaussian Error Linear Unit (GELU) function. Clinical and imaging data from MS patients make up a large dataset that is used in this research. The data is prepped for analysis using a variety of preprocessing techniques, including as feature extraction, data cleaning, and normalization. Alongside deep learning models like Convolutional Neural Networks (CNNs) are reviewed. Our results indicate that AI models, particularly those utilizing deep learning techniques, significantly outperform traditional methods in predicting MS. By utilizing GERF, it achieved lowest mean loss (0.1397) and highest mean validation accuracy (94.46^%) of MS predictions. GERF can directly replace Sigmoid, ReLU, tanh, Swish, Nip una, Serf, and GELU in deep neural network architectures. The study highlights how AI has the power to transform MS prediction and diagnosis, opening the door to more individualized treatment plans and improved patient outcomes.