To evaluate a newly developed algorithm for metal artifact reduction (MAR) in Computed Tomography (CT). A projection interpolation algorithm for MAR with threshold-based metal segmentation was developed. First, the algorithm was tested with a simulated hip phantom. On demand, the presence of metallic inserts was simulated, representing total hip endoprostheses. Second, CT data of 20 patient with total hip endoprosthesis were reconstructed with and without application of the MAR algorithm. Image quality was independently assessed by 2 experienced radiologists using a qualitative score. The results of the in vitro study were evaluated with the Student's t test. Results of the in vivo study were analyzed using a repeated-measure analysis of variance. Applying the MAR algorithm the phantom study showed no significant difference between images with and without simulated metal contributions. The patient study revealed improved image quality using the MAR algorithm. Results were statistically significant for fat (P = 0.0097), vessels (P = 0.0091), and bone (P = 0.0005). Improvement of the image quality for muscle was not statistically significant (P = 0.0287). A new algorithm for metal artifact reduction was successfully introduced into clinical routine. The algorithm led to a robust reduction of metal artifacts. The MAR algorithm may serve for an improvement in image quality in patients with metallic implants.
Purpose: To quantify differences in regional myocardial perfusion in coronary artery stenosis by the use of dual source computed tomography (DSCT) in an animal model. Material and Methods: In 5 pigs, an 80% stenosis of the left anterior descending artery was successfully induced by partial balloon occlusion (ischemia group). Five animals served as control group. All animals underwent contrast enhanced whole heart DSCT (Definition Flash, Siemens, Germany) perfusion imaging using a prototype electrocardiogram -triggered dynamic scan mode. Imaging was performed at rest as well as under stress conditions during continuous infusion of adenosine (240 mg/kg/min). For contrast enhancement 60 mL Iopromide 300 (Ultravist 300, Bayer-Schering Pharma, Berlin, Germany) were injected at a rate of 6 mL/s. Myocardial blood flow (MBF), first pass distribution volume, and intravascular blood volume were volumetrically quantified. Results: In the control group MBF increased significantly from 98.2 mL/100 mL/min to 134.0 mL/100 mL/min if adenosine was administered (P = 0.0153). There were no significant differences in the perfusion parameters comparing the control and ischemia group at rest. In the ischemia group MBF under stress was 74.0 ± 21.9 mL/100 mL/min in the poststenotic myocardium and 117.4 ± 18.6 mL/100 mL/min in the remaining normal myocardium (P = 0.0024). Conclusion: DSCT permits quantitative whole heart perfusion imaging. As this technique is able to show the hemodynamic effect of high grade coronary artery stenosis, it exceeds the present key limitation of cardiac computed tomography, which currently only allows a morphologic assessment of coronary artery stenosis.
The aim of this study was to systematically compare the ability to assess the coronary artery lumen in the presence of coronary artery stents in multislice spiral CT (MSCT). Ten different coronary artery stents were examined with 4- and 16-detector row MSCT scanners. For image reconstruction, a standard and a dedicated convolution kernel for coronary artery stent visualization were used. Images were analyzed regarding lumen visibility, intraluminal attenuation, and artifacts outside the stent lumen. Results were compared using repeated-measure analysis of variance. Depending on stent type, scanner hardware, and convolution kernel, artificial lumen narrowing ranged from 20% to 100%. The convolution kernel had the most significant influence on the visibility of the stent lumen. Artificial lumen narrowing and intraluminal attenuation changes decreased significantly using the dedicated convolution kernel. In general, most severe artifacts were caused by gold or gold-coated stents. Independent of the scanner hardware or dedicated convolution kernels, routine evaluation of most coronary artery stents is not yet feasible using MSCT.