Ung Jang1, Dosik Hwang1
1School of Electrical &
MR venography (MRv) has an important role in diagnosis of venous diseases. Unlike x-ray or CT venography, MRv produces venographic images without any ionizing radiation. Obtaining venographic image with low noise and high contrast is clinically important for diagnosis of vascular diseases based on MRv. It has been reported that the best venography can be acquired at TE=28ms for veins in parallel to the static field of 3T while much longer TE is needed for the veins not parallel to the static field. However, such a long TE would result in lower signal-to-noise ratio (SNR) in MRv image since the signals from veins and other tissues drastically decay with time. Therefore, reducing noise at longer TE is important in order to acquire a clinically meaningful venography. Employing conventional spatial filters such as low-pass filter, median filter, and anisotropic diffusion filter are typical denosing methods. Although such spatial filters reduce noise, they also smooth out the details of images such as small structures and edges, or introduce an artificial appearance such as stair-casing artifact due to nonlinear process. These degradations of image quality resulted from interference of neighbor pixels. The aim of this study is to obtain high-quality venography at longer TE without introducing blurring effects or any artificial appearance, by reducing noise on temporal domain of 3D multi-echo datasets.