Segmentation of CT Brain Stroke Image using Marker Controlled Watershed

In this paper, an algorithm is proposed to detect and segment ischemic stroke from CT brain images. Firstly, our proposed method starts by a preprocessing step contains skull bone striping and text removal from CT images, then the images are enhanced using median filter and histogram equalization. Next the watershed segmentation and Marker Controlled watershed methods is applied to detect the ischemic stroke. The experimental results show that Marker controlled watershed is better than the watershed due to over segmentation caused by the noise of the CT image. The over segmentation problem was resolved and succeeded to detect and segment the infarcted regions in the CT Ischemic stroke image that will help the non-radiologists identify the stroke visually.