An MRI Based Algorithm for Detecting Multiple Sclerosis

Early diagnosis of brain disorders can significantly reduce the devastating consequences of these disorders. Physician’s diagnostic capabilities and diagnosis time can be improved using computerized diagnosis techniques. Magnetic Resonance (MR) images are used for diagnosing multiple sclerosis, which is a disease that occurs when the immune system eats away at the protective covering of nerves. MR images segmentation is a complicated task due to the variability in the lesion’s shape, location and patients’ anatomy. This study proposes a new computerized diagnosis technique for detecting brain disorders based on features extracted from MR images. Data of 121 cases were used, including healthy and patients with brain disorders. The cases were classified into normal and abnormal, with abnormal representing brain disorders cases. The abnormal cases were fed into a classifier to identify brain disorders. Classification accuracies in the two stages were 82.7% and 70%, respectively; indicating a significant improvement over methods found in literature. The automated structure of the proposed algorithm is suitable for use in hospitals at low cost.