Mediated Spatiotemporal Registration of Cardiac DCE-MRI and Coronary MR Angiography

Patient-specific correlation of perfusion defects to coronary arteries which are responsible for blood supply in the affected territories has the potential to improve accuracy of diagnosis and intervention planning. The correlation of coronary arteries and perfusion defects can be established through registration of 3D angiography and 2D perfusion datasets. However, cardiac cycle phase mismatch between angiography and perfusion datasets causes the vast variation of the myocardium shape in these datasets and precludes the use of the standard methods for 2D/3D registration. Research published to date does not provide a reliable method of registering 2D perfusion and 3D angiography datasets. This paper presents an approach for non-rigid registration of 2D perfusion and 3D angiography data; the solution to the registration problem relies on the use of the 4D (3D + time) cine series to mediate non-rigid registration of perfusion and angiography datasets. The approach is evaluated on clinical data and its performance is compared with previously suggested methods for perfusion and angiography registration. The evaluation results show the utility of the presented method in the context of perfusion analysis while highlighting its applicability to other areas of cardiac image analysis.