Cardiac perfusion imaging with single photon emission computer tomography (SPECT) is a widely used diagnostic test for assessing coronary artery disease. Typically, the images are quantified and compared with a database of normal values. However, attenuation of the relatively low-energy photons emitted by TI-201 or Tc-99m (the two principal agents used) distorts the images obtained. This distortion may increase the standard deviation of the normal database, thereby decreasing the sensitivity and specificity of the test. This study investigated the degree to which attenuation correction might decrease the normal database variability, thereby increasing the sensitivity and specificity of the test. To accomplish this correction, simulations were performed, based on PET images. In these simulations, actual cardiac positron emission tomography (PET) images were forced to have """"""""normal"""""""" uptake of tracer, by setting all pixels within the myocardium to a constant value. These images were then projected through the measured, true attenuation map, and reconstructed using the standard SPECT filtered back-projection methods. The SPECT-reconstructed images could then be compared with the known uniform background. Preliminary data were presented orally in June 1995. Data have now been obtained to examine this effect separately for men and for women (who are purported to have significantly different attenuation patterns because of breast tissue). The full study is being incorporated into Mr. Lan's doctoral thesis, which was to be completed by September 30, 1996.
