Automatic Brain Hemorrhage Segmentation and Classification in CTscan Images

Brain hemorrhage detection and classification is amajor help to physicians to rescue patients in an early stage. Inthis paper, we have tried to introduce an automatic detection andclassification method to improve and accelerate the process ofphysicians’ decision-making. To achieve this purpose, at first wehave used a simple and effective segmentation method to detectand separate the hemorrhage regions from other parts of thebrain, and then we have extracted a number of features fromeach detected hemorrhage region. We selected some ofconvenient features by using a Genetic Algorithm (GA)-basedfeature selection algorithm. Eventually, we have classified thedifferent types of hemorrhages. Our algorithm is evaluated on aperfect set of CT-scan images and the segmentation accuracy forthree major types of hemorrhages (EDH, ICH and SDH)obtained 96.22%, 95.14% and 90.04%, respectively. In theclassification step, multilayer neural network could be moresuccessful than the KNN classifier because of its higher accuracy(93.3%). Finally, we achieved the accuracy rate of more than90% for the detection and classification of brain hemorrhages

Keywords:

brain hemorrhage , segmentation , GA , neural network , KNN classifier