Abstract
We developed an image processing method to estimate an instantaneous 2-dimensional blood flow distribution using cineangiograms (CAG). This method will be useful for in vivo study of blood flow. Two-dimensional blood flow velocity was estimated from the flow of contrast medium between two sequential CAG frames. Each frame was divided into overlapping block areas whose size was determined by consideration of the maximum flow velocity. Then we made a cost function including smoothness constraint term. The apparent interframe movement of the contrast medium between blocks was determined by minimizing the cost function. A 2-dimensional blood flow velocity could be, consequently, estimated. By introducing overlapping block areas, a spatially dense distribution of blood flow velocity was obtained. We tested this method with an experiment using a pulsatile flow through a curved pipe. A time sequence of 2-dimensional flow velocities in the curved pipe was estimated with this method, and a time sequence of spatially averaged flow velocities in the pipe was calculated. The calculated average flow velocity showed excellent agreement with electromagnetic flow measurements (γ=0.960, n=47). Furthermore, the estimated flow velocity profile in the curved pipe showed that the maximum flow velocity shifted towards the outer wall. This result is consistent with a visual inspection of the CAG frames. To evaluate not only averaged flow velocity but also flow velocity profile, we also made a computer simulation study. We simulated steady fluid flow in a straight pipe. By this method, parabolic flow velocity profile in the straight pipe could be estimated within small enough error. Finally we applied this method to clinical data. The application to an abdominal aorta with an aneurysm showed a vortex flow in the abdominal aneurysm. The application to a left ventricle showed not only the blood flow in the left ventricle but also the left ventricular wall motion. These results were consistent with visual inspection of CAG frames.
