Mitral valve rigid registration using 2D echocardiography and cardiac computed tomography

This study proposed a registration framework to register 2D echocardiography images with cardiac CT volume. The registration facilitates the fusion of CT and echocardiography to aid the diagnosis of mitral valve diseases. The image registration framework consists of two major steps: temporal synchronization and spatial registration. Temporal resolution utilizes the ECG data to identify frames that are at similar cardiac cycle phases as the CT volume. Spatial registration is an intensity-based normalized mutual information (NMI) method applied with pattern search optimization algorithm to produce interpolated cardiac CT image that matches the echocardiography image. Our proposed registration method has been applied on the four chamber view of the heart. Dice coefficient and Hausdorff distance for the intermodality assessments were 0.86±0.05 and 1.50±0.02 mm respectively. This accuracy is comparable to the gold standard manual registration by experts. Without the use of optical tracking devices, the outcomes indicate the applicability of this technique for effective registration of echocardiography with cardiac CT volume.