Purpose: We compared two methods for measuring myocardial wall thickening from nuclear medicine perfusion scans. The first method uses the percent change in peak activity, and the second method models a profile measured across the myocardium.
Method: Mathematical simulations of the myocardium were used. In addition, images with PET or SPECT resolution were created from real MR images. Known amounts of noise were then added.
Results: The percent peak thickening (% PT) is nonlinear with true percent thickening, especially for PET resolutions [7 mm full width at half-maximum (FWHM)]. For the peak method, low levels of noise (10%) introduced an error of 8%PT for PET and of 16%PT for SPECT. Additional smoothing reduced these errors. For the fitted model, at 10% noise, the error in thickening was large: 2.3 mm for PET and 7.8 mm for SPECT.
Conclusion: The fitted model works well only with good resolution and low noise (e.g., 7 mm FWHM and 10%). The peak method is also sensitive to noise, especially for poorer resolutions. Additional smoothing gives more reliable results for the peak method but not the fitted method. The peak method is therefore the more generally reliable, but even this method may only allow classification of myocardial thickening into broad categories.