2D and 3D multimodality cardiac imagery: application for the coronarography/SPECT/PET-CT

Coronarography and tomo-scintigraphy (SPECT, Single Photon Emission Tomography) are two imaging techniques used broadly for the diagnosis of cardiovascular diseases. The first modality consists of X-ray image sequences visualizing each, in a same plane, the coronary arteries located on the front and the back side of the heart. The X-ray images give anatomical information relating to the arterial tree and highlight eventual artery narrowing (stenoses). The SPECT modality (nuclear imaging) provide a 3-dimensional (3D) representation of the myocardial volume perfusion. This functional information authorizes the visualization of myocardial regions suffering from irrigation defaults. The aim of the presented work is to superimpose (in the 3D space) the functional and anatomical information in order to establish the visual link between arterial lesions and their consequence in terms of irrigation defaults. In the 3D representation chosen to facilitate the diagnosis, the structure of a schematic arterial tree and the stenoses are placed onto the perfusion volume. The initial data consist of a list of points representative for the arterial tree (start and end points of arterial segments, bifurcations, stenoses, etc) and marked by coronary-graphs on the X-ray images of the different incidences. The perfusion volume is then projected under the incidences of the coronary-graphic images. A registration algorithm superimposing the X-ray images and the corresponding SPECT projections provides the parameters of the geometrical transformations bringing the points marked in the X rays images in equivalent positions in the 2-dimensional SPECT images. A 3D reconstruction algorithm is then used to place the arterial points and the stenoses on the perfusion volume and build a schematic tree acting as landmark for the clinician. A 28 patient database was used to realize 40 3D superimposition of anatomical-functional data. These reconstructions have shown that the 3D representation is precise enough for the establishment of the visual relationship between stenoses and perfusion defaults. (author)