S-Space College of Engineering/Engineering Practice School (공과대학/대학원) Dept. of Electrical and Computer Engineering (전기·정보공학부) Theses (Ph.D. / Sc.D._전기·정보공학부)
Blood Flow Visualization using Flowline Extraction and Opacity Modulation based on Vascular Structure Analysis
혈관 구조 분석 기반 혈류선 추출과 불투명도 변조를 이용한 혈류 가시화 기법
- 공과대학 전기·컴퓨터공학부
- Issue Date
- 서울대학교 대학원
- Blood flow visualization; Integration based flow visualization; Automatic seeding; Flowline extraction; Opacity modulation; Vascular structure
- 학위논문 (박사)-- 서울대학교 대학원 : 전기·컴퓨터공학부, 2016. 2. 신영길.
- With recent advances in acquisition and simulation of blood flow data, blood flow visualization has been widely used in medical imaging for the diagnosis and treatment of pathological vessels. The integral line based method has been most commonly employed to depict hemodynamic data because it exhibits a long term flow behavior useful for flow analysis. This method generates integral lines to be used as a basis for graphical representation by tracing the trajectory of a massless particle released on the vector field through a numerical integration. However, there are several unsolved problems when this previous method is applied to thin curved vascular structures. The first one is to locate a seeding plane, which is manually performed in the existing methods, thus yielding inconsistent visual results. The second one is the early termination of a line integration due to locally reversed flow and narrow tubular structure, which results in short flowlines comparing with the vessel length. And the last one is the line occlusion caused by the dense depiction of flowlines. Additionally, in blood flow visualization for clinical uses, it is essential to apparently exhibit abnormal flow relevant to vessel diseases. In this paper, we present an enhanced method that overcomes problems related to the integration based flow visualization and depicts hemodynamics in a more informative way for assisting the diagnosis process. Using the fact that blood flow passes through the inlet or outlet but is blocked by vessel wall, we firstly identify the vessel inlet or outlet by the orthogonality metric between flow velocity vector and vessel surface normal vector. Then, we generate seed points on the detected inlet or outlet by Poisson disk sampling. Therefore, we can achieve the automatic seeding that leads to a consistent and faster flow depiction by skipping the manual location of a seeding plane to initiate the line integration. In addition, we resolve the early terminated line integration by applying the tracing direction adaptively based on flow direction at each seed point and by performing the additional seeding near the terminated location. This solution enables to yield length-extended flowlines, which contribute to faithful flow visualization. Based on the observation that blood flow usually follows the vessel track if there is no obstacle or leak in the middle of a passage, we define the representative flowline for each branch by the vessel centerline. Then, we render flowlines by assigning the opacity according to their shape similarity with the vessel centerline so that flowlines similar to the vessel centerline are shown transparently, while different ones opaquely. Accordingly, our opacity modulation method enables flowlines with unusual flow pattern to appear more noticeable, while minimizing visual clutter and line occlusion. Finally, we introduce HSV (hue, saturation, value) color coding to simultaneously exhibit flow attributes such as local speed and residence time. This color coding gives a more realistic fading effect on the older particles or line segments by attenuating the saturation according to the residence time. Hence, it supports users in comprehending intuitively multiple information at once. Experimental results show that our technique is well suitable to depict blood flow in vascular structures.