| With the rapid development of the medical image technology, cardiac image data scanned by multi-slice spiral CT plays an important role in the clinical diagnosis of cardiovascular. How to conveniently and quickly assist doctors getting valuable vascular diagnostic information correctly from huge coronary artery CTA data is always a big challenge for medical image researchers. The core technology of coronary artery computer-aided diagnosis and realization mainly includes cardiovascular segmentation and centerline extraction in complicated data, curved planar reformation (CPR) served for diagnosis, quantitative detection and qualitative analysis for vascular lesions and system architecture design and realization etc.Cardiovascular segmentation is the first step for diagnosing the cardiovascular disease. It is also a prerequisite for the following diagnosis of3D reconstruction, quantitative detection and qualitative analysis etc. The segmentation is difficult in the coronary artery CAD system caused by the high complexity and random of coronary artery CTA data. Then a method based on features of tubular strucutrue and the relativity of inter-slice is proposed in this paper to solve this problem, which utilizes the position information of ascending aorta and the structure features of coronary artery. The experimental result on coronary artery CTA data indicates that the proposed method can segment coronary arteries automatically and effectively, then reduce the interactive times and improve doctor’s work efficiency.Vascular centerline is widely used in vascular visualization, vascular lesion detection and analysis and other related fields. Because of the complexity of vascular structure and curvature, the extracted centerline is usually inaccurate. Thus, an automatic centerline extraction method is proposed in this paper based on the vascular topology structure feature and the idea of direction tracking. The experimental result indicates the proposed method can effectively solve the problem of inaccurate centerline extraction caused by lager curvature changes and amounts of sub-branch.CPR is a visualization technology that assists doctors observing the tubule’s inter-structure and shape trend, which has an important value for vascular clinical diagnosis.Firstly current CPR technology is discussed and analyzed, then to overcome the shortcoming such as the insensitive to tiny lesions, a rotating multi-angle CPR is proposed in this paper to better assist doctors observing.The quantitative detection and qualitative analysis of cardiovascular disease is the core goal and fundamental value of coronary artery CAD system, which can save time and reduce work intensity in the course of diagnosis. The current stenosis detection method based on vascular cross-section areas is firstly researched and discussed. To improve the high error rate and insensitivity of current method, an innovative detection method based on morphologic feature is proposed by using the shape features of vascular cross-section. The experimental result indicates the proposed method has much higher sensitity, which can effectively assist doctors detecting and analyzing vascular diseases.The core technology of coronary artery CAD needs to be imbedded into a stable, secure, customizable and extensible platform, then can be transferred into the industry. Thus, the coronary artery CAD platform system is also be researched in this paper. It can not only maximum assist cardiovascular diagnosis, but also improve the reuse and expansion capacity of each core technology and realize the industrial transfer by using a three-layer architecture of computer software system and object-oriented design methods. The clinical application result proves that the CAD system based on the researched core technology can effectively assist doctors diagnosing the cardiovascular disease under the CTA mode, improve the diagnosis accuracy and work efficiency, which has a great clinical signifaciance and the value of practical application. |
PDF Full Text Request