Multi Segmentation Method for Hemorraghic Detection

The beneficial of technology of Computer Tomography (CT) Scan has been recognized useful to examine all human organs, particularly for human's brain. CT Scan can detect stroke on hemorrhagic type stroke because it shows very clear different between white's matter and gray's matter. The segmented area is handy in providing human's brain information that shows which is the bleeding area and which is not bleeding. The intensity of pixels characterized by the maximum number of the segmented regions can be used to differentiate bleeding from healthy tissue. The focus of the study was to calculate the area of hemorrhage, the volume of hemorrhage, the value of Hounsfield Unit (HU). Differences in HU value in each patient signify the difference between bleeding due to blockage (infarction) or due to rupture of blood vessels of the brain. Image data information is obtained by conducting ground truth on a CT Scan image indicated by the hemorrhagic stroke with the dominant white matter feature in the brain. Ground truth performed presents an evaluation of several segmentation methods such as contour segmentation, adaptive otsu, watershed segmentation and clustering using Fuzzy C Mean method using image of CT Scan using 20 patient data indicated by the hemorrhagic stroke. Segmentation is done on 2D CT Scan data in axial position. The evaluation method was performed to obtain the accuracy value of bleeding image reconstruction by comparing with ground truth done by the radiologist.