Coronary CT angiography using iterative reconstruction 
vs. filtered back projection: evaluation of image quality

To compare image quality of iterative reconstruction algorithm(IRIS) vs. standard filtered back projection(FBP) reconstruction in CT Coronary Angiography (CTCA).

Thirty-four consecutive patients underwent CTCA for suspected or known CAD with Dual-Source CT (DSCT-Flash, Siemens). All datasets were reconstructed with 0.75/0.4 and 0.6/0.4 mm slice thickness/increment, using three standard FBP kernels (B26-B30-B46) and three comparable IRIS algorithms (I26-I30-I46). Vascular attenuation and noise were measured. CT vascular attenuation values [HU] were measured in: ascending aorta (Ao), right (RCA) and left (LCA) coronary artery, respectively. Signal-to-noise (SNR) and contrast-to-noise (CNR) ratio were calculated. A p-value<0.05 was considered significant.

There was no significant difference between the vascular attenuation values measured with FBP (Ao:458HU, RCA:448HU, LAD:444HU) and IRIS (Ao:456HU, RCA:446HU, LAD:442HU). Difference in noise was significant between FBP (24±SD) and IRIS (19±SD) (r=0.34;p<0.05). Lowest noise was found for IRIS using 0.6 mm (17HU). IRIS provided a SNR and CNR significantly higher with increasing kernel sharpness. SNR was 33.3±25.1, 77.3±51.7, 37.2±36.6, 64.4±59.2, while CNR was 25.32±19.8, 58.0±36.0, 28.6±23.5, 47.6±47.3 for 0.75B, 0.75I, 0.6B and 0.6I, respectively. IRIS showed an improvement in SNR of 57% and 56% for 0.75 mm and 0.6 mm, respectively, and an improvement in CNR of 42% and 40% for 0.75 mm and 0.6 mm.

In CTCA, iterative reconstructions provide a significant higher image quality compared with the conventional FBP reconstructions. (www.actabiomedica.it).