Applying J-optimal channelized quadratic observers (J-CQO) to a clinical imaging study for ovarian cancer detection

The identification of blood vessels is an important image processing technique in several medical imaging applications. Detection of blood vessels in computed tomography (CT) images is essential in diagnosis and surgical planning. Computed tomography (CT) is a common imaging modality used for blood vessels imaging. It can provide a high quality images for both diagnosis and treatment purposes. However, it is known that CT is a source of high radiation dose especially for cases required a follow up scans in short time. Reducing the dose is knows to produce statistical noise and artifacts that significantly reduce the image quality. In this paper, we propose a method for three-dimensional (3D) blood vessel detection from CT images reconstructed from small number of projection views. The proposed method is implemented in two stages. First, is a preprocessing stage to reduce noise. In the second stage, we use skeletonization method to extract the centerline of 3D object. Then, we use the blood vessels skeleton to construct a graph model for 3D vascular network. Finally, we use the graph model to detect blood vessels. The proposed method is tested using 3D blood vessel-like phantom and good results are achieved.