Midline Shift Detection in Brain CT Images by Segmentation using Gaussian Mixture Model

Abstract

In this paper we propose an image segmentation algorithm based on Gaussian Mixture model combined with template match. Accurate detection of midline shift of CT brain scans is vital for diagnosis and treatment of Traumatic Brain Injuries (TBI). Automatic processing of these CT brain scans could speed up the decision making process, lower the cost of healthcare, and reduce the chance of human error. In this paper, we focus on automatic processing of CT brain images to segment and identify the ventricular systems. The segmentation of ventricles provides quantitative measures on the changes of ventricles. Initially the ideal midline is detected to calibrate the CT position and also for reference later. Segmentation of the CT images is the important step of this application. The specific aim of this method is to extract ventricles from brain CT images. In the method, different types of brain tissue, of which the ventricles form the region of interest, are segmented using multiple Gaussian mixtures.  Expectation Maximization (EM) method is used to train the GMM. Ventricular tissue has to be then detected in the segmented regions using template matching. The algorithms have then to be evaluated against a dataset of brain CT images captured from both normal and TBI cases. The proposed GMM-based method will allow accurate segmentation of ventricles required for detection of the shift in the midline shift in brain.