Fast Brain Abnormality Detection Method for Magnetic Resonance Images (MRI) of Human Head Scans Using K-Means Clustering Technique

Abstract

This paper proposes a rapid method to classify the brain MRI slices into normal or abnormal using brain extraction algorithm (BEA), K-means and knowledge based techniques. BEA is used to extract the brain part from the original magnetic resonance images (MRI) of head scans. K-means is a simple and quicker segmentation process used to segment the brain into known brain regions, like white matter (WM), gray matter (GM) and cerebro-spinal fluid (CSF). Any abnormalities of brain usually affect the normal brain tissues (BT). At times, their intensity characteristics are identical to CSF class. This knowledge is used to analyze the segmented classes of brain by K-Means and thus identify the abnormal slices and location of abnormality within the slices. Experiments were done with datasets collected from medical schools. The results were compared with existing method. The proposed work took only 2 s to produce the results where as the existing requires 12 s per brain extracted slices. The proposed method never produced wrong classification but sometimes missed the abnormal slices. But the existing method had mixed possibilities. This proposed method could be used as a preprocessing technique in brain related studies and thus saves radiologist’s time, increases accuracy and yield of diagnosis.