Improving the image quality of contrast-enhanced MR angiography by automated image registration: a prospective study in peripheral arterial disease of the lower extremities.

OBJECTIVE If a patient has moved during digital subtraction angiography (DSA), manual pixel shift can improve the image quality. This study investigated whether such image registration can also improve the quality of contrast-enhanced magnetic resonance angiography (MRA) in patients with peripheral arterial disease of the lower extremities. MATERIALS AND METHODS 404 leg MRAs of patients likely to have peripheral artery disease were included in this prospective study. The standard non-registered MRAs were compared to automatically linear, affine and warp registered MRAs by four image quality parameters, including the vessel detection probability (VDP) in maximum intensity projection (MIP) images and contrast-to-noise ratios (CNR). The different registration types were compared by analysis of variance. RESULTS All studied image quality parameters showed similar trends. Generally, registration improved the leg MRA quality significantly (P<0.05). The 12% of lower legs with a body shift of 1 mm or more showed the highest gain in image quality when using linear registration instead of no registration, with an average VDP gain of 20-49%. Warp registration improved the image quality slightly further. CONCLUSION Automated image registration can improve the MRA image quality especially in the lower legs, which is comparable to the effect of pixel shift in DSA.