Comparison of radial and Cartesian imaging techniques for MR coronary angiography

Magnetic resonance coronary angiography (MRCA) has traditionally been performed using a Cartesian k-space data acquisition scheme. Radial k-space sampling is known to be less sensitive to motion artifacts. Thus, potential improvements may be achieved with radial k-space data acquisition using steady state free precession (SSFP) techniques. We directly compared SSFP three-dimensional (3D)-MRCA using radial and Cartesian data acquisition.

Forty-four consecutive patients with suspected coronary artery disease underwent free-breathing, navigator-corrected MRCA of the left or right coronary artery using SSFP (TR/TE/flip angle: 4.5 ms/2.3 ms/90 degrees) with radial and again with Cartesian k-space filling. Quantitative MRCA was performed with a dedicated multiplanar reformatting software to determine: visual score for image quality (low=1, high=4), vessel sharpness, visible vessel length, number of visible side branches, and average vessel diameter. Diagnostic accuracy for detection of > or = 50% coronary artery stenosis was calculated in comparison to invasive X-ray angiography.

Radial data acquisition resulted in a significant (p<0.01) increase in vessel sharpness (55.6+/-7.2% vs. 45.9+/-7.0%) but a decrease in average vessel diameter (2.6+/-0.5 mm vs. 3.0+/-0.4 mm), number of visible side branches (2.1+/-1.1 vs. 3.0+/-1.7) and number of assessable coronary artery segments (66% vs. 73%) compared to Cartesian approach. There was no significant difference regarding the diagnostic accuracy (80.8% vs. 83.9%), the visual score (2.6+/-0.9 vs. 3.0+/-0.9) and the visible vessel length (92.1+/-36.0 mm vs. 99.9+/-32.4 mm).

MRCA with radial k-space sampling appears to be on a par with Cartesian approach with respect to the diagnostic performance in an unselected patient population. Nevertheless, with current implementations, radial sampling is inferior to Cartesian sampling regarding the visualization of side branches despite better vessel sharpness.