Automatic Segmentation of Multiple Sclerosis Lesions in Brain MR Images Using Ensemble Machine Learning

Multiple Sclerosis (MS) is one of the most hurtful diseases that affect the brain and destroys permanently brain cells. The bad impact of MS affects human vision, balance, muscle control, and daily activity. Magnetic Resonance Imaging (MRI) is a popular medical tool used to detect many diseases. Manual human detection consumes time and depends on human views. Therefore, automatic models are used instead of manual segmentation models. However, using automatic segmentation to detect MS lesions is very challenging, it must be accurate. Many research efforts have carried out to automate the diagnosis process of MS lesions using machine learning techniques. Research work in this area, especially in large data, faces many challenges like time consumption, data acquisition, memory insufficiency, and low accuracy. Another important challenge is related to the training data set which is, in most cases, unbalanced. Imbalanced datasets are a common classification problem caused due to data types when the number of observations per class is not equally distributed. In our study, we consider these problems by introducing a new methodology that uses a hybrid machine learning model. Two-dimension discrete wavelet transform (2D-DWT) and textural features are used to extract local features from MR images. We propose two different hybrid models: Ensemble Support Vector Machine (ESVM) and Ensemble Decision Tree (EDT). We detect MS twice using both models and compare the results. Results show that the two models give the same accuracy. Considering the imbalanced data challenge, the hybrid model is considered amongst the top performing solutions.