Study of myocardial glucose metabolism in rats with PET using wavelet analysis techniques

The parametric images are attractive in PET imaging since they provide a global view of the imaged tissue including healthy and diseased structures. However these images are hampered by the noise due to the limited injected dose of the tracer. In this work we compare myocardial metabolic rates of glucose (MMRG) in rats obtained in the spatial and in the wavelet domains with the usual graphical method. The images were decomposed in the wavelet domain and all correlated pixels in approximations and details images were considered. Mean MMRG obtained from regions-of-interest drawn on parametric images in both methods showed values 15% higher while the variance was about 18% lower in the wavelet images. In conclusion, the wavelet filtering process allowed to enhance the image quality and to reduce the variance in the parametric images while preserving the spatial resolution.