Real-time matching of angiographies with in situ heart image sequences

Coronary angiography has become an important tool in modern medicine by providing an additional modality for heart disease diagnostics. Even though the recorded images (or image sequences) are digitally available, their analysis is very elementary in nature in most cases: the surgeon studies the angiograms before the operation, and operates without having them available in the operation room. It is up to the surgeon's memory to accurately remember the relevant details like locations of stenosis or occlusion. To improve both the analysis and the on-site availability of angiograms, but also to drastically increase the usefulness of the angiogram as a diagnostical modality, we present a novel approach which combines several advanced techniques into a single tool allowing real-time augmentation of in Situ beating heart image sequences with the corresponding angiogram. The software toolset we developed so far contains the removal of specularities from the glossy heart surface and the reconstruction of the original image colors, color based coronary artery segmentation in beating heart image sequences, angiogram segmentation and symbolic representation, OpenGL based augmentation and visualization, Mass-Spring-NURBS-based soft tissue simulation. Future work will focus on developing robust matching techniques and on integrating the software tools into an operation room suitable hardware setup.