Wahle A, Prause GPM, DeJong SC, Sonka M:


Determination of the Absolute Axial Orientation of Intracoronary Ultrasound Images in Fusion with Biplane Angiography.

Computers in Cardiology 1998, Cleveland OH, IEEE Press

Volume 25, Page 153-156, 1998

Preliminary description of a new algorithm for estimation of the absolute orientation of IVUS frames in fusion with biplane angiography (Continued) (Paper) (Links)

Abstract: While intravascular ultrasound (IVUS) gains increasing importance, assessment of spatial structures still lacks a geometrically correct 3-D reconstruction. The IVUS frames are usually stacked up to form a straight vessel, neither considering any curvature nor the axial twisting of the catheter during the pullback. Quantification of this simplified data inevitably results in significantly distorted values. To solve this problem, our approach assigns the results from IVUS segmentation to their geometrically correct spatial locations and orientations. One of the major tasks in this reconstruction is the determination of the axial rotation of the IVUS frames. The relative twist between adjacent IVUS frames can be calculated analytically from the 3-D catheter path using the Frenet-Serret rules. The question remains what the absolute axial orientation of the resulting frame set is. We are using the bending behavior of the imaging catheter as a reference, which is expected to fall in the location of minimum energy within the vessel. The path of the catheter is reconstructed from biplane angiography, and the location of each IVUS frame can be derived from the time function of the pullback. The extracted catheter path is then used as a guide line for the detection of the (contrast-dye filled) inner lumen within the biplane angiograms. The lumen center line is calculated from the center points of the 3-D ellipses resulting from the geometrical reconstruction of the projected lumen cross sections. For each frame location, a 3-D vector can be determined that shows the deviation of the center line to the path of the catheter. This difference is primarily caused by bending and must show its effect in the IVUS images as well. The vessel contours are segmented from the IVUS images and mapped into the 3-D space using the reconstructed location, an initial orientation, and the analytically determined relative twist. In each frame, the centroid of the lumen is calculated from the lumen contour. The error in absolute rotation is derived from the weighted difference angles between the out-of-center vectors extracted from angiographic vs. IVUS images, and is then minimized for an optimal fit over all frames.