Segmentation of Brain Stroke Lesions using Marker-based Algorithms in CT images

The computed tomography has huge role in the assessment of the hemorrhagic lesions of the brain. Physicians widely use CT to delineate the size and magnitude of the bleeding. In the medical image processing, the separation and detection of the objects is very crucial issue. The water-based segmentation (subdivision) is an approach that use to detect the closely contact margins tissues within the images. Manual outlining of the stroke in CT images considers as subjective operation that takes long time with less accuracy. In this study, the lesions were detected firstly and followed by Contrast augmentation and Segmentation. The suggested technique was evaluated to endorse its achievability and efficiency. These techniques attained 0.97 + 0.01, 0.98 + 0.02 and 0.991 + 0.01 (P = 0.001) for sensitivity, specificity and operating curve analysis, respectively. The analysis of the results images showed that the proposed approach is effective in detecting of the smaller lesions which might missed by using other segmentation methods.