Improved determination of biplane imaging geometry and 3D coronary arterial tree from two views

The authors propose a new technique for estimation of 3D biplane imaging geometry based on two perspective projection angiograms acquired at arbitrary orientations. With the calculated imaging geometry, the 3D coronary arterial structures can then be reconstructed. Given 8 or more corresponding object points in both views, a linear algorithm is first applied to obtain an initial estimate of the imaging geometry in terms of a rotation matrix [R] and a unit translation vector t/spl I.oarr/. The solution provided by the linear algorithm is then used as an initial estimate for the second step, in which constrained nonlinear optimization with uncertain prior information of the imaging system is employed to complete an optimal estimate of the imaging geometry. The accuracy of this method was evaluated by using a computer-simulated coronary arterial tree.<>