An automated segmentation of brain MRI for detection of normal tissues using improved machine learning approach

Due to an increased need for efficient and objective evaluation of large amounts of data, MRI-based medical image analysis is gaining attention in recent times. The goal is to simplify an image into something that is more meaningful and making it easier to analyze. The aim of medical image segmentation in brain MRI is to separate the region of interest from the background after denoising and skull removal. Accurate segmentation of normal and abnormal tissues is still a challenge for researchers. In this paper, we propose a fully automated segmentation of normal tissues viz., white matter (WM), gray matter (GM) and cerebro spinal fluid (CSF) from brain MRI using an improved machine learning approach that uses Neuro-fuzzy as classifier. The segmentation is carried out using gradient method and orthogonal polynomial transform. The performance of our method is assessed with metrics such as false positive rate (FPR), false negative rate (FNR), specificity, sensitivity and accuracy. Also, the entire procedure is developed as a graphical user interface (GUI) which results in automated classification and segmentation.