Desktop version

Home arrow Health

Abdominal Blood Vessel Segmentation

Abdominal blood vessel segmentation is an important task in CA. The basic framework of the segmentation process can be summarized as follows: (a) simple thresholding, (b) region growing, and (c) employment of a vessel enhancement filter. Because the density values of abdominal blood vessels and other organs are similar, intravenous contrast is used to enhance the vessels. Relatively large blood vessels have very high image contrast and higher CT values; segmentation of these blood vessels can be performed using simple thresholding techniques with some connected components analysis.

Thresholding-Based Vessel Extraction This method extracts abdominal blood vessel regions from 3D abdominal CT images. As stated before, contrast-enhanced CT acquisition is performed to achieve higher density in the blood vessels.

Figure 3.91 shows an example of abdominal contrast-enhanced CT images, which were taken in the arterial phase. Because abdominal regions are highly contrasted areas, it is easy to segment these regions using a simple thresholding technique followed by connected components analysis, which removes some bone areas.

Example of 3D-rendered view of abdominal contrast-enhanced CT image

Fig. 3.91 Example of 3D-rendered view of abdominal contrast-enhanced CT image

Region Growing The process extracts abdominal blood vessels by tracing high- intensity regions on contrast-enhanced CT images. The first step, that of continuous growing of regions from a given starting point, often uses spherical structure elements. Typical growing conditions are based on intensity values. Because abdominal blood vessels have a branching structure, growing conditions considering such branching are also developed.

Blood Vessel Enhancement Filter The diameters of abdominal blood vessels visible on CT vary from 30 mm in size to submillimeter. Because of the partial volume effect, the intensity of small blood vessels is lower. Also, absolute intensity values of small blood vessels become smaller than those of large blood vessels. Thresholding or region-growing methods may fail to detect abdominal blood vessels because of such phenomena. This is the reason for using a blood vessel enhancement filter. Second-order differential intensity analysis is a popular technique in enhancing blood vessels on 3D CT images. Hessian-based analysis is widely used in the field of medical image analysis [81, 243]. Blood vessel enhancement filtering can be performed by the method shown in Sect.

First a Hessian matrix H is obtained at a target point p and eigenvalues Ab A2, A3 (0 > Ai > A2 > A3) are computed for a predefined scale. The blood vessel regions are extracted as a set of voxels that satisfy

There is always an array of diameters of vessels as they branch. Because the Hessian-based approach of blood vessel region extraction is quite sensitive to the diameter of a blood vessel, a multi-scale approach changing the ะพ in Hessian matrix computation is performed to enable appropriate extraction. Figure 3.92 shows an example of blood vessel regions extracted by Hessian-based analysis.

Example of blood vessel extraction based on Hessian analysis

Fig. 3.92 Example of blood vessel extraction based on Hessian analysis

S. Hanaoka et al.


< Prev   CONTENTS   Source   Next >

Related topics