Iterative model reconstruction improves image quality, reduces dose in cCTA

Knowledge-based iterative model reconstruction (IMR) reduces intravascular noise on coronary computed tomography angiography (cCTA) by 86 to 88 percent and betters image quality at radiation exposure levels 80 percent below standard technique, according to a study published in the June issue of Academic Radiology.

Although clinical CT image reconstruction was previously based on filtered back projection (FBP), hybrid methods that combine FBP with iterative reconstruction have been developed to reduce noise. These newer techniques not only further reduce noise image, but also offer the potential for dose savings and improved image quality, wrote lead author Ethan J. Halpern, MD, of Thomas Jefferson University in Philadelphia, and colleagues.

Increased clinical use of cCTA has started an introduction of dose reduction techniques, including iterative reconstruction, to minimize radiation dose. Halpern and colleagues evaluated knowledge-based IMR’s ability to improve quality and reduce radiation dose in cCTA by reviewing diastolic phase reconstructions of cCTA exams and additional systolic phase reconstruction ascertained with reduced tube current using tube current modulation (TCM).

The researchers evaluated 45 cCTA studies, which included 25 studies that were performed with an 80 percent systolic dose reduction using TCM. Each study was reconstructed with FBP, hybrid iterative reconstruction (iDose4) and IMR in a diastolic phase. Halpern et al gathered additional systolic phase reconstructions for TCM studies.

After mean pixel attenuation value and standard deviation were measured in the left ventricle and left main coronary artery, two independent reviewers offered subjective scores on a five-point scale for definitions of contours of small coronary arteries, coronary calcifications, noncalcified plaque and overall diagnostic confidence for the presence or absence of stenosis.

Pixel density was not significantly different among FBP, iDose4 and IMR. Noise amplitude in the left main coronary artery was reduced by a factor of 1.3 from FBP to iDose4 in diastolic phase images. It was reduced by a factor of 2.6 from iDose4 to IMR. Noise amplitude in the left main coronary artery was reduced by a factor of 2.3 from FBP to iDose4 and by a factor of 3 from iDose4 to IMR in systolic phase TCM images.

The four subjective image quality scores were significantly better with IMR in comparison to iDose4 and FBP. The researchers found that the reduction in image noise amplitude and improvement in image quality scores were greatest among obese patients.

“Given the noticeable improvement in the definition of small vessels (<3 mm), it is possible that IMR will be most useful to improve visualization of small coronary vessels in large patients,” wrote the authors.