Classification of hematomas in brain CT images using neural network

Hematoma is common in traumatic brain injuries. An automatic detection and classification system helps doctors in analyzing the medical images. CT scan is the preferred method in traumatic brain injuries due to little cost, extensive availability, fast scanning and superior contrast. This paper deals with automated system to detect and classify the type of hematomas using artificial neural network algorithm for CT images of different patients. The methodology comprised of four phases, first preprocessing performed on the brain CT images, second histogram based centroids initialization for K-means clustering algorithm to segment the image in different clusters based on the intensity values of pixels. Third phase consists of features extraction from segmented image. In fourth phase, artificial neural network has been created and trained according to the features extracted from the image Trained artificial neural network (ANN) classifies the types of hematoma according to their features.