Abstract
Mapping of myocardial blood flow (MBF) with first‐pass perfusion imaging is becoming an important tool in the study of coronary artery disease. In this study a wavelet‐based denoising method was developed to improve the accuracy of pixel‐by‐pixel MBF maps. We performed an in vivo study in five stenotic dogs with 70% stenosis in the left coronary arteries. First‐pass perfusion imaging sessions were performed by administering the intravascular contrast agent Gadomer at rest and during dipyridamole‐induced vasodilation. Color microspheres (MS) were injected into the dogs to measure MBF at the same time. After denoising was performed, the signal‐to‐noise ratio (SNR) of the first‐pass perfusion image improved by approximately 180%, whereas spatial variation of MBF maps decreased 38%. It was also found that the correlation of MBFs measured by MRI with the MS method indicates a significant improvement with the denoising method (R^2^ increased from 0.24 to 0.78, P < .001). This suggests that the wavelet denoising method may be an effective way to increase the accuracy of pixel‐by‐pixel MBF quantification and reduce spatial variation, and may be applicable to other forms of noise‐sensitive image analysis. Magn Reson Med, 2006. © 2006 Wiley‐Liss, Inc.