Automatic Extraction of quasi-Synchronous Views from Rotational Angiographic Sequence without ECG-Data

In this contribution we present an automatic method to extract quasisynchronous views from a sequence of monoplane rotational coronary angiograms. Based on the work of Blondel et al. we build the image of horizontal integrals, which contains information about the vertical motion of the coronary tree. The maxima of this motion correspond to end-distole states. We extract a curve of vertical motion and perform a spectral analysis to extract local maxima. We perform a spatial analysis to detect irregularities in the cardiac cycle, in order to extract just the regular views. Finally we present experiments on several angiographic sequences and compare the results to available electrocardiogram (ECG-) signal.