Abstract
Our attempt consists of modelling the heart left ventricle at stress and rest situation, using the myocardial scintigraphic data and focuses on how to demonstrate differences in obtained 3D stress/rest images. 70 cardiac patients had completed myocardium tests by Tc-99m tetrofosmin and a GE-Starcam - 4000 SPECT gamma - camera. SPECT (Single Photon Emission Computed Tomography) slices were created and used. The myocardial perfusion was estimated by comparing those slices and the suspicion of an ischemia was indicated. 3D myocardium images were reconstructed by GE Volumetrix software in the GE Xeleris processing system by FBP reconstruction method, Hanning frequency 0.8 filter and a ramp filter and transferred in a Dicom format. The Dicom file, for each patient and each phase is imported to MATLAB 7.8 (R2009a). A series of isocontour surfaces were studied, in order to identify the appropriate threshold value, which isolates the myocardium surface from the rest area of the image. Based on the previously calculated threshold value, the myocardium volume was evaluated and be reconstructed in a 3D image. The possible difference relating to the rest and stress data of the 3D images, in voxels, was calculated, using MATLAB image processing analysis; the quantification and analysis of differences was followed. We tried to determine an index of quantification and define the global quantitative defect size as a fraction of the myocardial volume area in 3D images that will give confidence in cardiac perfusion efficiency recognition by SPECT. © 2010 IEEE.