Abstract: An approach for fusing X-ray angiogram and intravascular ultrasound (IVUS) images to accurately reconstruct vessels based on compensatory disadvantages of both imaging modalities is proposed. Firstly, the pullback path of IVUS catheter is reconstructed from angiogram pair acquired at the origin of the path. Then, vessel wall boundaries in cross-section of lumen are extracted from IVUS images with a semi-automatic snake-based method. Lastly, the relative changes in orientations of IVUS frames are determined with a Frenet-Serret frame-based model. Absolute orientation of each IVUS frame along the pullback path is obtained by using a non-iterative statistical optimization algorithm. Consequently, reconstruction of vessels based on fusion of two types of data is implemented. Experimental results with clinical data demonstrate the validity. Possible errors and computation cost are also discussed.

Key words: computer application, 3D reconstruction, image fusion, X-ray angiography, intravascular ultrasound (IVUS)