A New Technique of Flow Voids Segmentation on MRI Image for Cerebrovascular Disease

Abstract

The flow voids is the condition occurs when the MRI image has lost its signal due to flow of bloods and other fluids such as cerebrospinal fluid (CSF) and urine. Generally, the MRI images particularly the vessels that contain vigorously flowing blood is seen low signal and this may reflect to vascular patency. Moreover, the manual delineation method to visually detect the flow voids is tedious and time consuming. Recently, an image processing technique such as watershed segmentation is most recommended technique to segment the MRI images of flow voids. A common watershed transformation used for segmentation is the marker-controlled segmentation, but the application of such method is limited particularly due to over-segmentation and sensitivity to the noise. Therefore, in order to overcome such limitations, this study is proposed a new scheme of improved technique to segment flow voids image based on watershed and k-means segmentation algorithms. The proposed technique that involves pre-processing process and the improved watershed segmentation algorithm is used to capture the flow voids in the MRI images. The performance of the proposed technique is measured by evaluating its accuracy to detect flow-voids and hence the results are compared to the golden standard results provided by manual delineation method. The proposed segmentation technique reveals that it is has highly suffice to reduce over-segmentation detection of flow voids in the MRI images with accuracy up to 90%. From the comparison results, it is also shows that the new proposed has potential to be used as pre-processing tools for radiologists in the future.