White Matter Hyperintensities Extraction Based T2-FLAIR MRI Using Non-Local Means Filter and Nearest Neighbor Algorithm

Abstract

Brain White matter appearing as hyperintensities on T2-FLAIR Magnetic Resonance Imaging (MRI) has the association with risk of stroke or dementia such as Alzheimer's diseases and vascular dementia. In many researches, WMH are also demonstrated that they can predict an increased risk of cerebrovascular diseases. WMH are counted as an intermediate marker to identify a new risk factor based on their quantitative measurement. In this paper, we propose a method to extract WMH areas from T2- FLAIR MRI in order to measure WMH automatically. The proposed method consists of two segmentation steps. In the first phase, the combining of k-means clustering with morphology techniques is applied to remove brain matter out of cranium from T2- FLAIR MRI input image. Then in the second segmentation phase, non-local means filter is applied to the extracted brain for image denoising and nearest neighbor algorithm is used to separate brain image into 3 different classes of WMH area, non-WMH area and background area. The collected information is extremely essential in monitoring disease progress. Therefore, it plays an important role in developing computer-aided diagnosis (CAD) systems for detecting many cerebrovascular diseases.