Research And Applications Of3D Vessel Segmentation Techniques Based On CT Images

Three dimensional vessel segmentation techniques are of great importance in medical image processing, which could not only help understand the vessel anatomical structure and blood flow status, but also make key factor in vessel disease diagnosis. Also, delicate surgical planning and accurate guidance plays an importance role in successful image guided intervention. For instance, segmentation of cystic artery and blood vessels from CT images are significant to preoperative planning of laparoscopic cholecystectomy. Visualization of the segmentation results can help clinicians to understand the patient’s cystic artery variation types, which can reduce the risk of destroy major blood vessels in operations.In order to solve the above segmentation application, a comprehensive research of vascular three dimensional segmentation techniques was done in this paper. Firstly, DICOM standard was studied. The meaning of CT value is clarified. In order to improve the display effect of CT images, transformation of window width-center technique was discussed. Secondly, in the scale space theory framework, the blood vessel tubular model and the vascular structure filtering method based on characteristic of tubular vascular structure were studied. Thirdly, two important three dimensional techniques-region growing and geometric active contour method were discussed. Region growing method were differentiate in merging criterion and two different criterion were studied. In addition, a quite different region growing method-directional region growing was also studied. In part of geometric active contour method, level set curve evolution model and numerical calculation were discussed. Based on the above research results, a new three dimensional region growing method based on vascular tubular structure was proposed and was verified on clinical medical images. By comparing the results to golden standards of manually segmentation data by experienced clinicians, sensitivity and specificity of this method were calculated. It can be found that the new method can effectively segment vessels, especially the small cystic artery vessel. Those accurate segmentation results created favorable conditions for subsequent three dimensional visualization work. Finally, visualization algorithm including surface rendering and volume rendering techniques were studied. A visualization program with user interface was developed using.Net platform with ActiViz.NET. After importing vessel and calot segmentation results, this program could give visualization result in different viewpoint, which provided clinicians great convenience in the preoperative observation of cystic artery variation.